Filip Pesek, Founder & CEO of DonnaPro, a European executive assistant agency for CEOs and founders, on the AI paradox
SHARE THIS STORY
It’s funny how AI is everywhere in finance, but nobody has extra time. Whatever sector you’re in, finance teams are fully embracing all that AI has to offer. But the impact on time seems minimal. We’re not seeing reduced workloads or even smaller teams. And the reason is simple: AI hasn’t removed the finance workload. It’s moved it. Rather than producing information, finance teams are now checking it. And few businesses have prepared for that.
The work automation hides
Automation can now handle all sorts of tasks, from drafting board commentary to spotting exceptions in a reconciliation. But it can’t reliably perform the final checks. It can’t compare final figures with source data. It can’t review the wording or manage approvals. And it certainly can’t take responsibility when something goes wrong. And that’s the difficulty finance teams are facing. Particularly in FinTech, where AI adoption has moved quickly.
While a payments company might use AI to prepare parts of its board pack, someone in finance still needs to check every number against the underlying management accounts. And it’s a scenario repeated across lending, auditing, neobanks, investment firms… For every request that AI can help with, a human has to be there for the final review. Even an automatically generated investor update needs to be checked. And it all takes time and people. That’s why a finance team can look highly automated while still apparently taking forever to get things done.
The pressure is loaded. Management expectations are high. And timeframes are reduced, in the expectation that AI can do it all. But despite their growing digital tool kit, finance staff remain overloaded and overstretched. Not because they’re working inefficiently, but because they’re trying to do their job properly, in an environment where the real work is rarely understood.
The finance checking problem
Of course, this isn’t unique to fintech. Research from the Bank of England and FCA found that 75% of UK financial services firms were already using AI in 2024, with another 10% planning to adopt it within three years. Yet just 2% of AI use cases were fully autonomous. The rest still involved people somewhere in the process. And that matters, particularly in finance, because there’s so little tolerance for error. Accuracy is everything. The numbers have to add up. Which is why people still play such a vital role.
The same research found that 62% of AI use cases in financial services are considered low materiality, compared with 16% considered high materiality. Showing that AI is generally only being used where the consequences of an error are small. So, the more a business stretches AI use out to more sensitive and important areas, the greater the need for effective human oversight becomes. And the more output AI generates, the more there is to review.
What being AI-trained should really mean
And that’s one of the problems businesses are facing with AI training. Far too often, being “AI-trained” means little more than that a staff member knows how to write better prompts. When what it should mean is that they understand which finance tasks are appropriate for automation in the first place.
It’s a lesson that really could be helpfully applied to every sector. But in finance, where so much sensitive data is involved, and people’s life savings can be at stake, that distinction matters even more. Before handing anything to an AI system, all team members need to ask whether it should be handling that information at all. Because when the system gets something wrong, you then have the complicated issue of accountability.
Another thing highlighted by the BofE’s research is that although AI has its uses when working independently, it always renders better results when people validate and improve what it produces. Underscoring the fact that AI’s role isn’t to replace people, but to support them, and that people are still needed to provide the essential control points that make AI so valuable. So, perhaps instead of obsessing over the time that AI is saving finance teams, the better focus would be on how AI output is being checked, who’s validating, and who’s giving the final approval. And all too often, the answer eventually comes back to the CFO, founder, or finance lead… Which creates its own problems.
When every workflow ends with one person, you don’t just create bottlenecks; you make that person indispensable. That means that while some phases of the work happen more quickly, the overall process is slower. And when, for whatever reason, that person is unreachable, everything is at risk of coming to a standstill.
That’s why the checking layer matters
When you build in the right support – an executive assistant who understands the business and knows which tasks can safely be automated – you start removing that operational blockage. They’re not there to make executive decisions, but to check data and make sure that things that need senior approval, reach the right person for the job. And that’s what AI processes need if you want them to work effectively.
The thing about AI is that it can do a lot, but just like a newly employed junior assistant, it needs oversight and support. Particularly in finance. And that’s where businesses are coming unstuck.
Manas Saha, Head of Financial Services UK at Publicis Sapient, on why firms that build client context, adviser judgement, past decisions and compliance outcomes – held in a form that both people and AI agents can use – will compound their advantage. Firms that don’t will be renting the same intelligence as everyone else
SHARE THIS STORY
For decades, wealth management has operated under a simple equation in which growth required proportionally more advisers, more operational capacity and more cost, so as assets under management increased, firms scaled through people, processes and organisational complexity. Artificial intelligence is now breaking that equation, and the speed at which it is happening has caught much of the industry off guard.
Earlier this year, the launch of Hazel, an AI-driven tax-planning tool from Altruist, triggered billions in stock selloffs across listed wealth managers as markets began to question whether the traditional fee model can survive in a world of near-zero-cost analytics. Whatever one makes of the scale of that reaction, it revealed something important about how investors now view the sector, because they are already repricing wealth firms for AI risk before most of those firms have worked out what AI will change inside their own operations.
The model will not be the moat
The harder truth behind that repricing is that the models themselves are heading towards commodity status. Within a couple of years every wealth manager will have access to a powerful foundation model, bought from the same small group of providers on broadly similar terms, which means the lasting advantage will sit in the layer each firm builds on top of it, made up of client context, adviser judgement, past decisions and compliance outcomes held in a form that both people and AI agents can use. Firms that build that layer will compound their advantage with every client interaction, while those that neglect it will find themselves renting the same intelligence as everyone else.
Scarcity still shapes the business
Traditional wealth firms have long operated under scarcity, with scarce expertise, scarce adviser capacity and scarce client touchpoints shaping how the business runs. Clients now expect 24/7 service and hyper-personalisation, while advisers are already overloaded with back-office work that leaves them less time for the client conversations that create value. Most firms continue to rely on workflows built around manual reviews, disconnected systems and information that sits across CRM platforms, portfolio management tools, planning software, emails and adviser notes, so important client context is often fragmented across the organisation rather than available where decisions are made. Advisers spend valuable time gathering information instead of interpreting it, reviews happen according to calendars rather than real-world conditions, and work moves between planning, portfolio management, compliance and operations through error-prone manual handoffs.
Adviser time is the lever
With demand growing and adviser supply dwindling, productivity has become a strategic lever, and if AI can free up even 30 to 50 per cent of adviser time, a firm can channel that capacity into new clients, deeper relationships or significant cost savings. Capturing that value requires AI to become more than another productivity tool bolted onto existing processes, because the first generation of AI, which focused on summarising meetings, drafting communications and automating routine tasks, left the underlying operating model largely untouched.
Three stages of autonomy
The next stage embeds AI much more deeply into business processes through AI agents that coordinate data sources, prepare recommendations and manage multi-step workflows under human supervision and guidance. For a regulated firm the question that matters most is where the boundaries sit, and in practice it helps to think in three stages.
In assistive automation, the adviser initiates the task, checks the output and remains fully responsible for the decision, with AI supporting discrete work such as summarising, extracting data or drafting content under basic usage policies and human review. With guided orchestration, AI coordinates across systems, proposes actions and prepares fuller outputs such as follow-ups or scenario analysis, while a person supervises the workflow, approves recommendations and manages exceptions within defined permissions, approval points and audit trails. A fully AI-orchestrated workflows, agents execute multi-step processes within agreed rules, monitoring portfolios, triggering actions and routing exceptions, while people set the boundaries, monitor outcomes and intervene wherever judgement is required.
Governance from the first day
That third stage is where governance becomes most demanding, and UK firms already have a framework that should shape their design from the very first day. Under the Senior Managers and Certification Regime, accountability for any activity that AI supports, recommends or executes remains with the firm and its senior managers, so every agentic workflow needs a named owner, clear audit trails, escalation routes, outcome testing, explainability, bias monitoring and board-level oversight capable of showing that AI is supporting better decisions and better client outcomes rather than simply faster processes. Governance designed this way becomes part of the operating model itself, and the ability to preserve context, explain recommendations, trace decisions and coordinate work within clearly defined boundaries is precisely what allows AI to operate safely at scale across regulated workflows.
The Intelligent Wealth Firm
In our recent white paper with PIMFA, we describe the destination as the Intelligent Wealth Firm, a continuously learning organisation in which human expertise and AI capabilities operate as an integrated system. A single client interaction may be informed by hundreds of AI-driven activities spanning research, planning, compliance, servicing and portfolio management, and every recommendation, decision and outcome feeds back into the firm’s memory, making the organisation progressively smarter over time. That memory captures historical recommendations, adviser actions, client interactions, compliance outcomes and business results, and over the years this experience becomes institutional knowledge that improves future decisions and creates an advantage that competitors, technology vendors and foundation model providers cannot easily replicate.
Context informs technology, so the data comes first. Most of what a wealth firm knows about a client sits in unstructured form: adviser notes, emails, meeting transcripts, provider documents, suitability reports, call recordings. Conventional systems cannot read it. AI can. Agents ingest that material, connect it to the structured records in the CRM and portfolio systems, and hold the result where the next adviser or the next agent can use it.
The gains show up in metrics that operations teams already track. First time right, the share of cases that clear onboarding, suitability or review without rework, should improve significantly when AI tools, ours included, work from information that is complete from the start. Every reworked case that disappears returns capacity to the back office. Some of that capacity lowers the cost of running the operation. The rest goes back to advisers and service teams, who get more time with clients.
Screening shows the effect clearly. A politically exposed person check can return 5,000 potential hits. An agent working with the right context, including the client’s history, source of wealth and earlier decisions, can narrow that to the ten that need a person’s judgement and record its reasoning on the rest. The time saved is direct, and it lands in the back office first.
Growth, retention and the client
Efficiency is half the case. The other half is revenue. Advisers with more time and better information can serve more clients, respond faster and see needs before a client raises them. Clients who feel known stay. A firm built around the client, not around its own systems, keeps more of the relationships it wins.
The obstacle is practical. Client data lives in many places, so the questions are how to bring it to one place, how to mine it quickly and how to consolidate the platforms underneath. We have invested significantly in all of our products to answer them. Sapient Bodhi builds and orchestrates agentic workflows with industry context-built in. Sapient Slingshot reads legacy code, documents the current-state by extracting business rules and dependencies and then migrates to a more modern codebase in an accelerated time, while taking much of the risk out by folding several platforms into fewer.
The enterprise context graph
Bringing the data together is the start. An enterprise context graph maps how clients, accounts, advisers, products, policies, decisions and outcomes relate to one another, and keeps that map current. It is the shared context layer behind our platforms, and it grows more accurate the more it is used. An agent preparing a client review can then see the holdings, and also the meeting where the client’s appetite for risk changed, the recommendation that followed and the compliance sign-off attached to it. The graph belongs to the firm. Models can be swapped as better ones arrive while what the firm has learned stays where it is.
Governing the context layer
A layer that holds client history, advice and compliance outcomes needs governing as carefully as the agents that draw on it. Access should follow permissions, so an agent sees only what the person it serves is entitled to see. Every piece of context needs a traceable source, so any recommendation can be followed back to the information behind it. Outputs need testing against client outcomes, and each workflow needs a named senior owner under the same regime. UK GDPR applies to the graph as it does to any other client record.
The first is to define decision rights for humans and AI
Establishing clear responsibilities for where AI can recommend, where it can execute and where human judgement remains essential, with governance, accountability and transparency designed into every workflow from the start. A useful exercise for any leadership team is to ask which activities in the business genuinely require human judgement and which persist simply because that is how they have always been done.
The second priority is to build the context layer
Connecting client information, adviser expertise, investment policies and historical decisions into a shared context that both advisers and AI agents can draw on when a decision needs to be made. This is where the durable advantage lies, and it is worth every firm asking what proprietary context, experience and decision-making knowledge it holds that a competitor, technology vendor or foundation model could not easily reproduce, since that knowledge is the raw material from which institutional memory is built. When we work with wealth firms on this through Sapient Bodhi, our enterprise agentic platform, most of the early effort goes into connecting that knowledge and building the governance around it, because the model a firm uses will keep evolving while its institutional memory keeps growing in value with every client interaction.
The third is to measure client outcomes in place of AI adoption
This sounds obvious but runs against how most AI programmes currently report their progress. It is tempting to track the number of use cases launched or the share of staff using a new tool, yet those figures say very little about whether the firm is getting better at what it does, whereas client results, decision quality, operational efficiency and improved governance are the measures that will matter to boards, to regulators assessing Consumer Duty and ultimately to clients themselves. They are also the measures that show which workflows deserve to be scaled across the organisation, so that every deployment strengthens the firm’s institutional memory and makes the whole system more capable with each client interaction.
Any wealth leader can start that work tomorrow by asking which decisions in their firm are still waiting for a meeting, a spreadsheet or a manual review, because each one of those is a place where the firm’s own memory could already be doing the work.
Errol Rodericks, Product & Solutions Marketing Director at Denodo, on why infrastructure, security, governance, resilience and accountable business authority remain essential
SHARE THIS STORY
For years, people have discussed sovereignty in financial services primarily in terms of geography. Where is the data stored? Which jurisdiction applies? Who operates the infrastructure? Those questions still matter. But they are no longer enough.
Banks and insurers now operate through an increasingly complex web of cloud providers, SaaS platforms, data ecosystems, third parties, AI models and legacy systems. Modern financial services could not function without these dependencies. The more important question is whether the institution remains in control.
Can it determine where sensitive information is used? Will it understand which critical services depend on which providers? Can it change a platform without rebuilding every data integration? Can it preserve consistent policies and business definitions as information crosses systems? And if a provider, region or legal basis suddenly becomes unavailable, can it continue serving customers within acceptable tolerances?
That leads to a different definition of sovereignty. Sovereignty is not technological self-sufficiency. It is controlled choice.cThe objective is not to remove dependency, but to ensure that critical dependency remains deliberate, governed and, where necessary, reversible.
The sovereignty problem is becoming an architecture problem
One of the biggest threats to institutional control is not simply fragmented data. It is fragmented control.
A typical financial institution has authoritative information distributed across core systems, SaaS applications, multiple clouds, data warehouses, lakehouses, APIs and external providers. Over time, integrations multiply, data is copied, pipelines proliferate, policies are implemented differently across platforms, and business definitions diverge. Each individual decision may appear reasonable. Collectively, they can make the institution increasingly difficult to govern.
The same customer may be represented differently across systems. A masking policy may exist on one copy but not another. An application can become tightly coupled to a provider-specific schema. During an incident, identifying which services, data and customers are affected may require manual reconciliation. The institution may still own its data and hold contracts with its providers, yet its practical ability to exercise control has diminished.
Risk accumulates at every hand-off through aggregation risk, policy drift, conflicting meaning and exit friction. The sovereignty problem therefore becomes architectural. How do you maintain consistent control across a distributed environment without forcing every workload into the same place? Geography remains one control. Architecture determines whether that control survives across the wider estate.
From data sovereignty to decision sovereignty
AI makes this question more consequential. When technology primarily stored and reported information, sovereignty could be discussed largely in terms of data, infrastructure and access. AI consumes information to interpret, recommend and increasingly act.
That introduces another question: who controls the context through which the institution is understood? Consider a fraud system assessing an unusual international payment. The transaction may genuinely look anomalous, and the model may be functioning perfectly. But perhaps the payment is associated with a corporate Treasury event that has already been authorised.
The data may be correct. The model may be correct. Yet without the relevant business context, the resulting action may still be wrong. The same issue appears elsewhere in financial services.
A lending decision requires more than a credit score. It may also require current exposure, affordability, product policy and delegated authority. An insurance claim cannot always be understood from the claim record alone: policy conditions, catastrophe events, customer circumstances, fraud indicators, documents and third-party information can all change the appropriate response. An AI customer-service agent may know the customer and product but still need to understand consent, entitlement, jurisdiction and what actions it is authorised to perform.
Sovereignty therefore has to extend beyond controlling where information resides to controlling how it is understood and how that understanding influences decisions. That is the emerging challenge of decision sovereignty: can the institution reconstruct what information was used, what it meant at that moment, which policies applied, which technology participated, what authority was delegated and why the resulting action occurred?
Institutional understanding should not belong to the model
This creates an important architectural question for enterprise AI. Financial institutions are unlikely to depend on one AI model forever. They will use combinations of commercial models, specialist models, internally developed models, agents and orchestration technologies. Those technologies will change rapidly.
Institutional understanding should not have to change with them. If the meaning of “customer exposure”, “vulnerable customer”, “suspicious transaction” or “eligible claim” is embedded independently inside every application, vector store or AI model, changing technology becomes much harder. Part of the institution’s business understanding has effectively become coupled to the technology consuming it.
A more sovereign architecture separates the two. Authoritative enterprise sources remain authoritative. A governed information and semantic layer can provide approved views, business definitions, relationships, lineage, access policies and current operational context. Applications, analytics and AI can then consume that governed context rather than reconstructing institutional meaning independently.
This leads to an important principle for AI-era sovereignty: Models should be replaceable. Institutional understanding should not be. The architecture in the underlying briefing makes this separation precise: enterprise sources remain authoritative and policy-bound, governed semantic context sits between those sources and AI, and model selection remains a separate policy decision.
Sovereignty does not mean zero-copy
The same nuance applies to data movement. It is tempting to equate sovereignty with never moving data. But that is neither practical nor always desirable. Some workloads require persistent history. Others need caching for performance or continuity. Some analytical workloads suit physical data platforms better.
The better question is whether movement is a governed decision rather than an integration default. Can a query run against the authoritative source when appropriate? If information must be cached, can the institution control where that cache resides, how long it persists and how it is protected? If a physical copy is required, is its purpose explicit and are retention, deletion and access obligations preserved?
Live federated access, controlled caching and physical pipelines can all be legitimate patterns. The appropriate choice depends on sensitivity, lawful location, freshness and latency requirements, source capacity, continuity requirements and economics. The principle is not “copy nothing.” Do not make another copy the automatic price of using information.
Policy must survive consumption
Control can also disappear between policy definition and actual use. An institution may have strong privacy, security and conduct policies, but if those controls are implemented differently across copies, applications and channels, policy is no longer consistently authoritative. The risk grows when information is dynamically assembled for analytics or AI.
A stronger architectural approach brings together identity context, approved semantic views, fine-grained controls and source-native permissions at the point information is accessed. As the briefing argues, policy is strongest when business meaning, access context and source controls agree.
That does not mean technology determines policy. Legal, compliance, risk and business functions still decide what is permissible. Technology’s role is to make those decisions enforceable and evidenced when information is actually used.
Exit clauses are not the same as exit capability
Another uncomfortable aspect of sovereignty is something financial institutions increasingly need to confront. A contract may contain excellent portability and termination provisions. That does not mean the institution can actually leave.
Applications may depend on proprietary schemas. Pipelines may embed provider-specific logic. Business definitions may exist only inside a platform. Identities, caches, networks and operational processes may all be intertwined with the provider.
True reversibility therefore requires more than contractual rights. It requires stable interfaces, reusable semantics, visibility of dependencies, replacement patterns and, critically, rehearsal.
As the briefing puts it, exit is a capability only after rehearsal. Contracts are necessary, but measured execution determines whether portability is real. That gives boards a more useful question. Do not ask only, “Do we have an exit clause?” Ask, “When did we last prove that we could exercise it?”
Controlled interoperability is the alternative to isolation
None of this means financial institutions should retreat from cloud, SaaS, external data, ecosystem partnerships or AI. Quite the opposite. Greater control can enable greater openness.
A bank that knows where its information comes from, understands how it is being used, can enforce access policies consistently, preserves its own business semantics and can change consuming technology is in a stronger position to collaborate.
This is why the future of sovereignty is not technological isolation. It is controlled interoperability. Institutions should be able to connect what they need without surrendering control of what they own.
That requires infrastructure controls, identity and cryptographic controls, cybersecurity, model governance, contractual protections, operational resilience, human authority and accountable business ownership. No data platform replaces those responsibilities.
A logical data and context layer can, however, address an important part of the problem: keeping distributed information connected, governed and understandable without requiring every consumer to create another proprietary representation of the institution.
The board question is changing
Financial-services sovereignty should ultimately be judged not by where a server is located or by a provider’s nationality, but by the institution’s ability to exercise control when conditions change.
Can it identify critical dependencies? Can it govern sensitive information wherever it is legitimately used? Can it preserve its business definitions and institutional context? Can it change providers, platforms or AI models without reconstructing its information environment? Can it continue operating when a dependency fails? And when an automated decision has consequences for a customer, can it explain what happened and remain accountable for the outcome? Those questions turn sovereignty from a political or technological abstraction into an operational capability.
Financial institutions will remain interdependent. They will continue to rely on global cloud providers, SaaS platforms, data ecosystems, and rapidly changing AI technologies. The objective is not independence from them, but ensuring dependency never quietly becomes a loss of control. That requires an architecture in which authoritative data can remain governed, institutional meaning can remain consistent, and applications, platforms and AI models can change without forcing the organisation to reconstruct its trusted information environment each time.
In conclusion
This is where a governed logical data and context layer plays an increasingly important role. By connecting distributed information without making another central copy the default, preserving semantics and lineage, enforcing governance at the point of access, and making trusted business context reusable across applications and AI, financial institutions can preserve control while continuing to innovate.
They do not make an institution sovereign. Infrastructure, security, governance, resilience and accountable business authority remain essential. But by helping separate institutional data and understanding from the technologies that happen to consume them today, they can help preserve something increasingly valuable as those technologies change: control of the institution’s data and understanding remains with the institution.
Tamsin Crossland, Principal AI Architect at Icon Solutions, on why ethical thinking, empathy and broader perspectives, leave organisations better equipped to build AI systems that are not only intelligent, but fair, resilient and trusted
SHARE THIS STORY
Artificial Intelligence (AI) has been making headlines as it moves from a future-facing concept to a modern reality. From fraud detection and credit scoring to customer service automation and risk modelling, AI systems are increasingly shaping real-world outcomes, particularly within the financial services industry. But as AI adoption accelerates across payments and technology, one question has become unavoidable: who is training and developing AI models, and whose perspectives are shaping their decisions?
AI is not neutral; it reflects the data it is trained on, as well as the assumptions and biases of the teams that design, test and deploy it. When those teams lack diversity, the risk of reinforcing existing biases increases. As a result, building ethical AI is as much a human challenge as it is a technical one.
Gender representation
Women remain significantly underrepresented within the technology and payments industries, a reality that often begins in school, before career choices are made. Early perceptions that computing is ‘a boy’s hobby’ continue to influence subject choices at school and university level, narrowing the pipeline of female talent entering technical and computing roles. While slow progress has been made, recruitment processes can still carry subtle gender bias, with many IT and technology placements being given directly to men over women. This is shaping who feels encouraged or excluded from pursuing careers in AI and advanced technologies.
Real change must begin earlier in the classroom. Inclusive education initiatives, visible role models and outreach programmes can actively encourage girls to pursue careers in STEM. At the industry level, companies should critically evaluate recruitment strategies and consider more ambitious gender diverse targets, such as aiming for a 50% female graduate intake. This is not about optics, but about building a workforce that better reflects the society AI systems ultimately serve.
Why diversity is critical in AI development
The case for gender diversity has become even more compelling when examining AI training and deployment. AI models are built on historical data, data that may already contain embedded biases. Without diverse perspectives in the initial phase, challenging assumptions and the biases used to create AI models, these biases can be amplified once the AI system is deployed.
Gender-diverse teams bring a range of different perspectives and backgrounds. For instance, women bring valuable and often underappreciated perspectives to the process. Emotional intelligence and empathy, frequently described as ‘soft skills’, are increasingly critical as AI systems move beyond efficiency tools and into decision-making roles that affect people’s lives. The ability to anticipate unintended consequences, question whether an outcome feels equitable, and consider how different groups may experience an automated decision is essential to responsible deployment.
Encouragingly, women are already leading important conversations within AI ethics forums and industry working groups, such as Women4Ethical AI, AI4ALL and Women in AI (WAI). They are contributing to debates around transparency, fairness, explainability and responsible use. However, discussion alone is not enough, and now it is time for action. The next step is embedding these principles into product development cycles, procurement standards and executive decision-making processes.
Ethical AI cannot be treated as a retrospective compliance exercise, but as an integral part of the initial development phase.
Shaping the future of ethical AI
As AI continues to shape the future of payments and technology, responsible representation must be viewed as a strategic imperative. Without gender diversity, the risk of reinforcing systemic bias grows. With diverse teams guided by ethical thinking, empathy and broader perspectives, organisations are better equipped to build AI systems that are not only intelligent, but fair, resilient and trusted.
Peter Pugh-Jones, EMEA Field CDO at Confluent, explains why as more AI projects move into production, that gap will become harder to ignore. The conversation will shift away from what AI could do and towards what organisations are actually able to support in practice
SHARE THIS STORY
AI is no longer an experiment in financial services. It’s already embedded in day-to-day operations.
Banks are rolling out AI-driven customer service, investing in automation, and exploring how generative and agentic systems can improve decision-making. Around 92% of global banks are already using AI in at least one core function, which suggests the industry is well on its way.
As these initiatives scale, however, a different reality starts to emerge.
What works in a pilot often becomes more complex, not to mention expensive, when rolled out across the business. In many cases, the issue is not the AI itself, but the environment it is being deployed into.
There is no ‘magic bullet’ for AI in financial services. Its impact is shaped by how well organisations can access, connect, and act on their data — increasingly through real-time approaches such as data streaming.
When AI meets reality
This is particularly visible in customer service.
AI-powered chatbots and virtual assistants are now widely used to handle routine queries, categorise transactions, and guide users through basic processes, and in the right conditions they can deliver real efficiency gains.
Scaling those systems, however, is a different challenge altogether. Despite the level of investment, many of these systems continue to rely on fragmented or outdated information, which sees the experience quickly starts to break down.
For example, customer context can be missing or inconsistent. Conversations don’t carry across channels. Employees are forced to double-check or correct outputs, removing much of the efficiency they were meant to deliver.
In a sector where trust matters, that’s difficult to ignore.
Legacy systems are still doing most of the damage
Frameworks like the EU AI Act are shaping how organisations deploy AI, particularly in high-impact use cases such as credit scoring, fraud detection, and customer risk assessment. Requirements around transparency, auditability, and data governance are raising the bar for how AI systems are built and monitored. And rightly so, given the importance of trust in financial services.
In practice, though, legacy infrastructure is often the bigger issue. When data is fragmented across systems or difficult to access in real time, it becomes far harder to evidence how decisions are made or ensure models are operating on complete, up-to-date information.
In many cases, it is these underlying data limitations — rather than regulation itself — that slow progress. With many banks reliant on systems built up over decades through mergers, upgrades, and workarounds, they struggle to create a single, reliable view of a customer or transaction — something AI depends on.
This also helps explain how FinTech startups can outmaneuver industry giants. Free of technical or technological baggage, they can build around modern data architectures from the start and bring new ideas to market more quickly.
AI can’t reach its potential, or in some cases even run, in a legacy environment. Sticking with the customer service example, one such example is personalisation.
Personalisation only works if the data is current
Personalisation has been a goal in financial services for years, but in many cases it has been closer to segmentation than true individualisation. If AI is going to change that, it needs a flow of data that can tell it exactly what’s happening in the moment.
Hyper-personalised experiences depend on real-time context — understanding a customer’s behaviour and situation as it evolves, often enabled by data streaming to ensure systems are working from the most up-to-date information. Without that, interactions quickly become disconnected.
A customer might start a query in an app, follow up by email, and then call support, only to repeat the same information each time. This is not a limitation of AI, but of how information moves between systems.
When that flow is in place, interactions become far more seamless, with context carrying across channels and experiences feeling more consistent.
AI should support people, not replace them
There is a lot of discussion about AI replacing human roles in financial services, but a fully automated model is unlikely.
AI is highly effective at handling repetitive, high-volume tasks, like triaging customer queries or processing transactions. But more complex interactions still require human judgement, empathy, and context — particularly in areas such as lending, investments, or financial advice.
Customer expectations also play a role. While digital channels continue to grow, many customers still value the option to speak to a person, particularly in more sensitive or high-stakes situations.
As a result, the most effective model is increasingly a hybrid one. AI handles routine processes and surfaces insights, while people focus on more complex and sensitive interactions — provided both are working from the same reliable, up-to-date information.
Early investment has focused on what AI can do. Attention is now turning to what is needed to make it work properly.
That starts with data. Organisations need to ensure data is connected, governed, and available when it is needed. Too often, this work is treated as something to address later, when in practice delaying it tends to create more complexity and cost over time.
There is also a cultural element to consider. In many organisations, established ways of working can slow progress just as much as technology. Addressing that is just as important as modernising systems, particularly when AI initiatives span multiple teams.
A question of readiness
Financial services organisations are being asked to move quickly while maintaining high levels of control and accountability. That is not easy.
The organisations that make progress will be those that focus less on AI as a standalone capability and more on the conditions that allow it to work. If those conditions are not in place, results will remain inconsistent.
As more AI projects move into production, that gap will become harder to ignore. The conversation will shift away from what AI could do and towards what organisations are actually able to support in practice.
In most cases, it comes down to a simple question: is the data ready?
Cien Solon, CEO & Co-Founder of LaunchLemonade, on why the future of financial services will be shaped by those who can balance AI innovation with responsibility
SHARE THIS STORY
Every industry is racing to implement artificial intelligence. In financial services and other heavily regulated sectors, that race has been cautious. From automated underwriting and fraud detection to hyper-personalised customer experiences, AI offers FinTech companies unprecedented opportunities to innovate and scale. Yet alongside this promise comes the growing weight of regulation.
Different approaches to AI governance are already taking shape. In Europe, the EU is introducing a structured, risk-tiered framework that sets clear obligations depending on how AI is used. The UK is pursuing a more flexible, principles-based approach, allowing individual regulators to apply guidance tailored to their sectors. The US has a more decentralised model, blending federal direction with a growing patchwork of state-level rules. For FinTech founders operating across borders, this diversity can feel complex and, at times, overwhelming.
With these frameworks emerging, many fintech founders are beginning to hesitate. Questions around compliance costs, legal complexity and potential liability are causing some to reconsider or delay AI adoption altogether. For smaller fintech firms and scale-ups, the concern feels especially acute. Is the risk worth the reward?
It is. And founders who treat regulation as a strategic advantage, rather than a barrier, will be the ones who pull ahead.
Regulation as a Trust Multiplier
In financial services, trust is everything. Customers entrust FinTech platforms with their most sensitive data and critical financial decisions. Investors and partners demand reliability, transparency and accountability. Regulators expect firms to operate with integrity and control.
Robust AI governance can enhance trust across all of these stakeholders. FinTech companies that proactively adopt transparent and responsible AI practices signal maturity beyond their size. They demonstrate operational discipline alongside technical capability.
This matters especially now, when AI systems can appear opaque or unpredictable. Clearly communicating how AI is used, what decisions it informs, what data it relies on, and how risks are mitigated allows fintechs to differentiate themselves in a crowded market.
Compliance, when done well, builds credibility. And credibility, in financial services, is a powerful competitive advantage.
The Myth of ‘Compliance Paralysis‘
One of the most common misconceptions among FinTech SMEs is that AI regulation requires building entirely new compliance infrastructures from scratch. This belief often leads to what I call “compliance paralysis,” a reluctance to act because the problem feels too big to start.
In reality, much of AI governance maps directly onto frameworks that fintech companies already use. Risk management, model validation, data protection, auditability. These are familiar disciplines. AI extends them into new domains, but the foundations are already there.
Take model risk management. The steps a compliance team follows to validate a human-led underwriting process can be adapted to validate an AI-driven one. The logic is the same. The documentation requirements are the same. What changes is the technology performing the task, and the speed at which it operates.
I see this with our own clients. Regulated businesses assume they need to start from zero when it comes to AI governance. When we walk them through what they already have in place, the gap between current practices and regulatory expectations is almost always smaller than they feared.
The Agility Advantage of FinTech SMEs
Large banks may have dedicated compliance teams and significant legal budgets, but they also carry the weight of legacy systems. Retrofitting AI governance into complex, decades-old infrastructure is both costly and time-consuming.
Fintech SMEs have the advantage of agility. They can design and implement responsible AI practices from the ground up, embedding governance directly into their products and processes from day one.
This is exactly how we built LaunchLemonade. Every AI agent on our platform operates within a governance framework by default. Audit trails, data handling rules, transparency requirements. They are baked into the architecture, because bolting them on later is always harder and more expensive.
This compliant approach delivers real advantages. Long-term costs stay lower because governance is structural rather than retrofitted. Adapting to evolving regulatory requirements becomes faster because the foundations are already solid. And AI-enabled products reach the market ready to meet governance standards, rather than needing months of remediation after launch.
Regulation, in this sense, levels the playing field. Larger incumbents may have more resources, but smaller firms can move faster and design smarter. Those that embrace this will outpace competitors who delay.
Practical Steps for Leaders
For finance leaders looking to navigate AI regulation without overwhelming their teams, the key is to focus on pragmatic, high-impact actions.
Start by classifying AI risk early. A chatbot handling customer queries carries a very different risk profile from an algorithm making credit decisions. Understanding where your use cases sit on the risk spectrum allows you to prioritise governance efforts where they actually matter.
Be transparent by default. Inform users when AI is being used, explain its role in decision-making, and provide avenues for human review where appropriate. Clear communication builds trust and reduces regulatory friction.
Leverage what you already have. Map AI governance onto your existing risk and compliance processes rather than building a parallel system. This reduces duplication and accelerates implementation.
Build lightweight governance structures. You do not need a large compliance team to implement effective oversight. Documenting model assumptions, maintaining audit trails and assigning internal accountability can go a long way. At LaunchLemonade, we see regulated SMEs achieve meaningful governance with lean teams precisely because they focus on these fundamentals.
And know when to bring in expert input. Legal and regulatory expertise is valuable, but it does not need to be engaged at every step. Focus external support on high-risk use cases or areas of genuine uncertainty.
The Cost of Standing Still
If financial SMEs retreat from AI because of regulatory anxiety, innovation will not stop. It will simply concentrate in the hands of large incumbents and major technology companies with the resources to navigate complex compliance landscapes.
That concentration has broader implications. Competition reduces. Innovation cycles slow. Diversity in financial products and services narrows. Opportunities to improve financial inclusion and accessibility are missed.
Fintech and finserve has historically been a driver of disruption, challenging traditional models and expanding access to financial services. AI represents the next wave of that transformation. If smaller players step back now, they risk ceding that ground entirely.
Regulation as an Enabler
The conversation around AI regulation in financial services needs a shift in perspective. Regulation creates the guardrails that allow new technologies to scale safely and sustainably. It builds the trust required for widespread adoption. And it ensures that innovation benefits broadly, across industries and communities, rather than being concentrated among those who can afford to figure it out alone.
For fintech and finserve founders, the real question is how to innovate responsibly and strategically. Those who embrace that question will find that regulation is part of the answer.
AI is too important an opportunity for FinTech to ignore. Regulatory frameworks may seem daunting, but they are neither insurmountable nor inherently restrictive. When approached thoughtfully, they strengthen trust, streamline operations and unlock new avenues for growth.
Financial SMEs are uniquely positioned to lead in this space. Their agility, focus and ability to embed governance from the outset give them a distinct edge over larger, slower-moving competitors. The future of financial services will be shaped by those who can balance innovation with responsibility. The firms that recognise this will define the age of AI, rather than simply surviving it.
Cien Solon is the CEO and Founder of LaunchLemonade, a secure governed platform of AI agents for regulated industries. An experienced AI transformation leader, Cien has been building AI-powered solutions since 2018 and working with generative AI since 2022. Her mission is to ensure that smaller businesses in regulated sectors can adopt AI confidently, without the enterprise price tag. LaunchLemonade was included in Entrepreneur UK’s Top 100 Startups to Watch, and Cien was shortlisted in the Technology category of the 2025 Investec Early-Stage Entrepreneur of the Year Awards.
Robbie Tilleard, GM EMEA at Lorikeet, on why you don’t need a year, a team of twenty engineers, or a bespoke model trained on your data to deploy an AI in CX
SHARE THIS STORY
Revolut launched its AI financial assistant in April this year. Starling launched one three weeks before that. An in-app AI financial assistant has become table stakes for FinTechs overnight.
The good news is you don’t need a year to get there. All you need is a plan and the right data.
Eight Years or Eight Days
Revolut’s AI assistant took a year from announcement to launch, and Starling describes their assistant as the culmination of eight years of work. Both were building the infrastructure before it existed off-the-shelf.
FinTechs we work with deploy an AI assistant in weeks, not months. They don’t build from scratch. They buy infrastructure that’s already solved the hard parts… Knowing when to escalate, handling ambiguity without hallucinating, navigating the tone of a conversation where someone has had their card blocked and is not in the mood for corporate language.
Sound familiar? It should. Every technology shift in financial services has looked like this for a period. Early movers build bespoke, the rest wait for the case studies, and then someone works out you don’t need to build when you can buy much faster. Then it simply comes down to having the right data in place, connecting your systems, and putting in place a clear implementation plan. Financial services knows the playbook. The difference is that the loop has simply gotten a lot faster with AI.
Why Not Wait Longer Then?
Customer expectations have moved, shaped by AI interactions people have outside financial services (support queries answered in seconds, refunds processed mid-conversation), and the gap between ChatGPT responses and waiting for a queued ticket to a human agent is becoming visible in churn data. The FinTechs that move first will set the bar for what customers expect from anyone in the same category.
There is also a regulatory dimension. In the UK, the FCA moved Consumer Duty into more active supervision this year. The review published in March 2025 found that vulnerable customers continue to receive worse outcomes than other customers, particularly where firms have primarily digital customer journeys, and that most firms have underestimated the depth of outcomes monitoring required.
The reaction to AI in many compliance teams looks a lot like the reaction to autonomous vehicles. Every time a Waymo crashes, the genuine safety improvements (fewer accidents overall, faster hazard detection, no fatigue) get less coverage than the single incident.
Yet AI is going to help Consumer Duty over time, not hinder it. Autonomy with the right guardrails produces more consistent outcomes than human handling at scale. An AI system can understand a customer’s problem immediately, resolve it on the spot where it can, and route to the right internal team where it can’t, without queue times, without agent fatigue, without twenty different people interpreting the same policy differently. This ensures the core customer care team remains focused and can prioritise the cases with the most need. And this isn’t a far off future for consumers. The infrastructure to build it already exists off the shelf.
The European Dimension
There are additional points to consider if you operate in Europe. The EU AI Act’s full compliance requirements for high-risk AI sees financial services explicitly on the list.
Being compliant means documented risk management, transparency mechanisms, human oversight baked into the architecture and a traceable record for every decision that the system makes. The cost of getting this wrong is high: up to €35 million or 7% of global revenue for the most serious breaches.
Most organisations aren’t close to achieving requirements. More than likely they’ve rolled out AI as an experiment (that’s good!) but without a plan to get it scaled in production (not so good). Systems built purely to retrieve and respond weren’t designed with auditability or explainability in mind and it isn’t something you can bolt on later.
If you operate in Europe and have not mapped your current customer experience AI stack against these requirements, the time is now.
What You Actually Need
You don’t need a year, a team of twenty engineers, or a bespoke model trained on your data. You need focused use cases, infrastructure that handles the compliance layer, and a team willing to run a tight pilot with clear success criteria and a plan to scale.
You can move now. The companies that don’t are still lacing up.
Errol Rodericks, Director of product marketing EMEA and LATAM at Denodo, on why the future of insurance will be defined by whether humans and AI agents have the right data, in the right form, to make the right decisions
SHARE THIS STORY
AI adoption across the UK insurance sector continues to accelerate, yet the impact is falling short of the industry’s expectations. The narrative is now shifting beyond systems that analyse data and generate recommendations towards agentic AI. Capable of taking action, whether that means initiating claims processes, identifying potential fraud in real time, refining underwriting decisions or triggering next-best actions. Many see this as the next major leap forward for insurance, and they may be correct, but not for the reasons most people think.
While AI development certainly doesn’t lack pace, most of the insurance industry remains constrained by a more fundamental issue. AI cannot make trusted business decisions until it first understands the business. That understanding depends on trusted, relevant data being delivered in the right context and at the right time. Without it, insurers will struggle to give AI systems meaningful autonomy, however advanced the models become.
This is not because insurers lack data or technology. Over the past decade, the industry has invested heavily in data lakes, advanced analytics and AI tools, and integration and data engineering pipelines. Yet despite these investments, many organisations continue to struggle to move AI beyond proof-of-concept projects and into day-to-day operations.
Meeting the Challenge
The challenge is not data availability. It is ensuring data is trusted, relevant and readily accessible when decisions need to be made.
Few industries experience this challenge quite like the insurance industry. Decision-making depends on information drawn from multiple policy, claims and customer systems, alongside external data sources such as telematics, weather and credit data. When that information is fragmented or delayed, even the most advanced AI systems are left working with an incomplete picture, making it difficult for AI to understand the wider business context behind every decision.
Agentic AI does not solve this problem. It exposes it.
Access Alone Is Not Enough
Many organisations respond by building a shared data foundation consisting of a unified layer where humans and AI agents can access the same information. While this is directionally right, it is incomplete. The challenge is not that organisations lack a shared data layer. They struggle to deliver the right version of data for each decision, at the moment it matters.
Insurance operates on multiple, decision-specific views of data, each with distinct requirements:
Claims decisions depend on real-time, enriched incident data.
Underwriting relies on forward-looking risk models and external signals.
Fraud detection requires cross-entity patterns and behavioural analysis.
Customer servicing depends on a simplified, current policyholder context.
These are not variations of the same dataset; they are purpose-built representations of data, shaped by different latency, governance, and semantic needs, which becomes even more critical with agentic AI. Different agents operate at different points in the decision lifecycle, and require different data, in different forms, at different times.
A shared data layer can improve access to information, but access alone does not guarantee better decisions. Context transforms data into business understanding, and business understanding is what enables trusted decisions.
Turning Data Into Action
This is where many AI strategies stall. Most architectures are designed to store, process, and analyse data, but not to activate it at the point of decision. There is a fundamental gap between data being available and data being usable within real-time workflows.
Agentic AI operates directly in this gap. Without access to live, governed, and contextually aligned data, agents operate with partial understanding, and their outputs become unreliable. This is why many AI initiatives remain stuck in experimentation.
To move forward, insurers need to rethink how data is delivered. Not as raw datasets or reports but as data products. A reusable, governed, and outcome-aligned data asset designed to support a specific decision or workflow is what’s needed. Instead of exposing raw data, insurers should deliver contextualised, decision-ready views, with embedded governance and policy controls, consistent business semantics, and real-time access to internal and external sources.
For example:
A claims data product unifying FNOL, policy data, repair estimates, and external signals.
A fraud data product combining claims history, network relationships, and behavioural indicators.
An underwriting data product integrating internal risk data with third-party enrichment.
These are not static datasets. They are decision-ready data assets, designed to deliver the right information, in the right context, for a specific business outcome.
Why Data Quality Matters Now More Than Ever
For agentic AI to deliver value, data must be live, governed at access, semantically consistent, and traceable. This is where a logical data layer becomes critical, not just as an integration approach, but as a way to connect distributed data in real time, apply governance dynamically, and deliver consistent, business-ready views across systems. This enables both humans and AI agents to act with confidence, without introducing further fragmentation.
The insurers that lead in 2026 will not be those with the most advanced models. They will be the ones that connect AI directly to business outcomes. That means starting with the outcome, such as reducing claims cycle time, improving fraud detection, increasing underwriting precision, or enhancing customer experience, and working backwards to define the decisions, data and systems required to support them.
This is how AI moves from experimentation to operational impact.
Where AI Initiatives Succeed or Fail
The next phase of AI in insurance will not be defined by advances in model performance. It will be determined by how well AI understands the business it is supporting. That requires trusted, relevant data, but it also needs the business context that allows AI to interpret that data correctly and make decisions insurers can trust.
Agentic AI accelerates this realisation. It makes clear that data must be trusted, contextual, available at the moment of decision, and aligned to outcomes. The goal is not simply to give AI more information. It is to help AI understand how the business works well enough to make confident, consistent decisions. Those who solve this will scale AI successfully, and those who do not will continue to pilot without transformation.
The future of insurance will not be defined by whether humans and AI agents share the same data. It will be defined by whether they have the right data, in the right form, to make the right decisions. That requires a shift from shared data to decision-ready data, from access to activation, and from experimentation to measurable outcomes. The real inflection point for AI in insurance will come when organisations enable AI to understand the business well enough to make decisions that customers, employees, and regulators can trust.
Hugh Scantlebury, CEO and Founder of Aqilla , on why the finance teams that benefit most from AI will likely be the ones that strike the right balance between automation and oversight
SHARE THIS STORY
At first glance, finance and accounting appear to be ideal environments for AI integration. The work is structured, rules-driven and built on numerical data, which AI can automate and process at scale and speed. As such, it’s ideal for much of the repetitive work — such as invoice capture, reconciliations, reporting, and anomaly detection. These tasks still consume so much time and reduce space and capacity for strategic and creative thinking.
But as confidence in AI grows across the sector, and as the technology is integrated into accounting and finance software, is there a risk that organisations once fearful of the technology may go to the other extreme? What might happen if they lean so deeply into AI that finance and accounting professionals become distant and removed from the numbers?
These questions matter because the decisions that sit behind the numbers are rarely driven by pure logic. Financial strategy is shaped by risk appetite, leadership judgement, organisational priorities and sometimes even internal politics.
So many entrepreneurial success stories include a fateful risk, gamble or moment of inspiration that defies established financial wisdom. A founder deciding whether to invest in growth or a charity balancing financial sustainability with its mission does not rely on numbers alone. That means a surprisingly high number of commercial decisions have a human dimension that AI cannot, and arguably should not, replace.
Balancing AI and Human Strengths
So how can organisations maintain the human intuition and instinct that sits behind so many corporate transactions while embracing AI? The answer is to use the technology to remove repetition rather than people from core financial processes. In practical terms, that means using automation to reduce cognitive load so people can focus on interpretation, creativity and decision-making — the human parts of finance and accounting.
That’s a sensible approach because poorly implemented automation accelerates errors and obscures decision-making processes. It can also cut humans out of the process exactly at the point where instincts and experience are most needed. In that situation, AI-enabled systems simply become faster at making poor decisions. And those poor decisions end up costing time as well as money.
Once that happens, the space, time and resources that finance and accounting leaders are trying to create for more strategic work will rapidly shrink. By contrast, when humans actively guide AI systems, review outputs and set boundaries, automation becomes a powerful extension of human capability rather than a substitute for it.
The Limits of Automation in Decision-Making
In finance and accounting, numbers often create a sense of objectivity and certainty. But financial reporting still involves interpretation, context and judgement. A technically correct output is not always the same thing as the right commercial decision for a business, its employees or its long-term strategy.
As such, it’s important that accounting and finance professionals at every level do not lose a grip on their data when engaging with AI and automation. Aside from losing an understanding of the systems and processes behind the outputs, there’s still an ethical responsibility to deploy AI in a way that preserves human oversight, authority and compliance in financial reporting. That’s because, for the first time, we’re asking technology, in the form of AI, to provide an opinion on our data — not just deliver the logic and the numbers.
For that reason, finance leaders still need visibility into how AI-generated outputs are reached, the ability to challenge them when necessary, and a clear understanding of the underlying data behind the results. That’s important because AI output is ultimately based on prediction, and prediction is not the same as established and quantifiable truth.
Keeping Score
One possible middle ground is confidence scoring and validation workflows. Rather than unquestioningly trusting every AI-generated output, organisations can introduce processes that flag lower-confidence results for human review before action is taken. That creates a more balanced relationship between automation and oversight, while also giving finance teams clearer visibility into how reliable or complete AI-generated outputs actually are. The goal should be confidence in AI-supported workflows, not unquestioning reliance.
That visibility, however, should not sit solely with grads and junior finance staff. They need to understand the manual calculations and processes sitting behind automated systems so they can properly challenge the results. But the same principle applies at senior levels too. Experience and seniority should not create distance from the underlying logic behind the numbers. If anything, AI makes that visibility even more important.
Otherwise, organisations risk creating the worst of both worlds: juniors who can’t challenge AI outputs because they haven’t learned the manual processes, and complacent seniors who know the manual systems and assume AI is following them. It means organisations can end up with more information at their fingertips than ever before, while simultaneously becoming more detached from the underlying data and logic behind it.
That risk becomes even more significant at senior levels, where financial decisions often carry wider operational, commercial and strategic consequences. The more detached leaders become from the logic behind the outputs, the greater the impact when reliability or credibility issues emerge within the data.
Conclusion
Productivity gains from AI and automation have the potential to create more space for higher-value thinking rather than remove people from the process. The information AI surfaces can ultimately help senior leaders make more creative and strategic decisions by revealing connections, patterns and insights that may previously have remained hidden within the data.
Leaders have always relied on summaries, dashboards, and reporting layers to help them make decisions. But AI dramatically widens the gap between decision-makers and the operational reality beneath the numbers. The danger is not simply inaccurate data. It’s overconfidence. When systems appear highly intelligent and highly efficient, organisations can gradually stop questioning how conclusions are reached in the first place.
Ultimately, organisations may wish to focus on people-led automation rather than handing complete control to systems. The goal should be confidence in AI-supported workflows, not unswerving reliance on automated outputs — with AI acting as an extension of human capability rather than a replacement for judgement.
The finance teams that benefit most from AI will likely be the ones that strike the right balance between automation and oversight. Hand over the repetition and the manual processing, but don’t lose visibility into the data itself. Because if organisations surrender that understanding completely, all the entrepreneurial instinct, creativity and commercial judgement in the world may no longer be enough to compensate for the decisions being made underneath them.
Simon Pamplin, CTO at Certes, on how Quantum computing will eventually force a reckoning with the cryptographic assumptions on which the financial system is built.
SHARE THIS STORY
Financial institutions hold some of the most sensitive and long-lived data of any sector. Customer identities, decades of transaction records, credit histories, and proprietary trading information must be kept confidential, not just today, but far into the future. The cryptographic systems guarding that data, however, were built for a different era; one in which quantum computing was purely theoretical. That era is ending.
Recent guidance from both the G7 Cyber Group and the UK’s National Cyber Security Centre has made the stakes explicit: financial organisations should complete their transition to post-quantum cryptography by 2035. For institutions managing sprawling, decades-old technology estates, that deadline is less reassuring than it might first appear. Ten years sounds like breathing room. In practice, for organisations of this complexity, it is barely enough.
Quantum Risk: The Threat Doesn’t Wait for the Technology
The most common mistake in discussions about quantum risk is treating it as a problem for tomorrow. The assumption is that until a quantum computer powerful enough to crack modern encryption actually exists, there is no immediate danger. That assumption is dangerously flawed.
Sophisticated adversaries are already operating on a longer time horizon. The strategy, widely referred to as “harvest now, decrypt later,” involves systematically collecting encrypted data today and storing it until quantum computing capabilities catch up. No quantum computer is needed to execute this phase of the attack, only patience and storage.
For the financial sector, the implications are particularly serious. The data being harvested now, customer records, transaction histories, and commercially sensitive communications, could have a useful life measured in decades. By the time it is decrypted, the individuals and organisations it concerns will still be very much affected by its exposure. This is not a theoretical future liability. It is a risk quietly accumulating in the background of every financial institution that has not yet begun to act on it.
Why Updating Financial Systems Is So Hard
The cryptographic methods underpinning most financial infrastructure today, primarily RSA and elliptic curve cryptography, derive their security from mathematical problems that are extraordinarily difficult for conventional computers to solve. Quantum computing threatens to render those problems tractable, potentially dismantling encryption that currently appears robust.
Responding to that threat, however, means working through some of the most technically complex environments in any industry. Core banking platforms were often built 20, 30, or 40 years ago. Cryptographic functions are frequently embedded deep within application logic, firmware, or hardware, in payment terminals, ATMs, and proprietary systems that were never designed with replaceability in mind.
Replacing or updating these components is not a software patch. It can mean recertifying devices, replacing physical infrastructure, or undertaking large-scale application rewrites, all while maintaining continuous service to customers and satisfying stringent regulatory requirements. Progress is possible, but it is measured in years, not quarters.
Rethinking Security from the Data Outward
The quantum threat is also accelerating a broader rethink of how financial institutions approach security architecture. Perimeter-based defences, firewalls, VPNs, and network segmentation have consistently shown their limitations when attackers can move laterally through systems using legitimate credentials or compromised access. Relying on the boundary to keep data safe has a poor track record.
A more resilient approach centres protection on the data itself. By applying strong controls directly to information as it moves across systems, organisations can maintain security even within legacy environments that cannot be immediately upgraded. Sensitive data remains protected regardless of where it travels or which network it traverses.
Alongside this, crypto agility, the ability to swap out cryptographic algorithms and rotate keys without overhauling entire systems, is becoming a fundamental requirement rather than a nice-to-have. The transition to post-quantum standards will not be a one-time event. Cryptographic best practice will continue to evolve, and institutions that have built flexibility into their architecture will be far better positioned to keep pace.
The Cost of Inaction
The financial sector has a well-established capacity to absorb and adapt to technological change. It has navigated the transition to digital banking, the rise of real-time payments, and successive waves of cybersecurity threats. The quantum challenge is different in one important respect: the timeline for harm is already running.
Regulatory expectations are being set now. Adversaries are collecting encrypted data now. And the internal processes required to modernise a major financial institution’s cryptographic infrastructure take years to complete. The gap between when action is needed and when it can realistically be delivered is not wide, and it is narrowing.
Preparation does not demand panic or the immediate replacement of every system. It demands a clear-eyed audit of where sensitive data lives, an honest assessment of which information carries long-term confidentiality requirements, and a commitment to building security architectures that can evolve as the threat landscape does.
Financial institutions that move early will not simply reduce their quantum exposure. They will emerge with stronger, more adaptable security postures and with the trust of customers, regulators, and partners intact.
Quantum computing will eventually force a reckoning with the cryptographic assumptions on which the financial system is built. The question is no longer whether that reckoning is coming. It is whether institutions will be ready when it arrives.
Chris Tredwell, Chief Operating Officer and Charis Thomas, Chief Product Officer at Aqilla, on why the question is no longer whether to adopt AI, but whether processes, governance structures and training pathways are ready for the workforce
SHARE THIS STORY
Have you ever got into an old car with a Gen-Zer? If they were driving, chances are you wouldn’t have got very far. A recent survey has found that 39% of 14–29-year-olds couldn’t identify an ignition key. Proof, if it were needed, that once technology advances, old ideas are quickly forgotten. This isn’t just happening in cars. The internet and social media have produced their own native generations – people who have never known a world without those technologies.
The same pattern is starting to emerge with AI. That means Gen Z and Millennials are about to experience a similar shift. The first wave of true AI natives will soon enter the workforce – a cohort that has never known a world without AI.
AI- The New Normal
People’s reactions will largely depend on their experience with AI. But one thing is certain: these graduate and entry-level employees won’t need to be convinced of its value. They’ve already seen what it can do, so if it’s missing, disbelief – or frustration – is likely to follow. It’s a bit like broadband. Here in the UK, it’s simply the standard we all expect. We don’t stop to think about how that connectivity reshaped our lives, helped us work from home or allowed us to stream high-definition media.
Many organisations are still in the early or experimental phases of AI adoption. They might be using the technology to automate basic email inbox management and take meeting minutes. Meanwhile, those further ahead of the curve are exploring more advanced tools and assessing where automation can be safely deployed, particularly for reporting and analysis.
But AI natives won’t see these use cases as experimental. In fact, they probably wouldn’t even refer to them as use cases. It’s just normal, like using a search engine rather than visiting a library to carry out research.
Prompting New Behaviour
Perhaps the biggest difference, however, is where organisations may integrate AI into their existing workflows, AI natives are more likely to structure work around it from the outset.
For them, work tends to start within an AI system, defining the objective clearly, setting constraints, and effectively “briefing” it, before iterating quickly and refining outputs as they go. For AI natives, this kind of prompt-based mindset isn’t a specialist skill; it’s simply how they approach tasks.
This is a fundamental shift. For AI natives, the question isn’t “Should we use AI here?” It’s “Why can’t I use it for this piece of work?” When their expectations collide with more cautious, process-led environments, friction is almost inevitable. Not because one approach is right and the other is wrong, but because both sides are starting from completely different assumptions.
Skills Transfer and Mentoring
But how does the need for AI natives to understand and work through basic manual processes coexist with intuitive prompt-based thinking? Should AI use come with experience-based restrictions in the finance sector? For example, do your three years first, and then you can use the tools.
It’s probably not what AI natives want to hear, but there is logic behind the approach. Learning the manual processes behind automation will enable new recruits to apply the necessary checks and balances to system outputs — putting them in a position to verify data rather than passively accept it.
Without taking this step, there’s a real risk that people will lose the ability to question the outputs they’re working with. That, in turn, has implications for how people are taught. Whether in an educational setting or on the job, that training needs to help AI natives understand the logic behind the systems they’ll be working with.
Lurking in the Shadows
If AI natives encounter friction when trying to use the technology, there’s a risk they’ll seek informal workarounds. Organisations have seen similar patterns before with personal devices or early cloud adoption. With AI, the risks are more focused on data, traceability, and accountability than on system access and security, though these remain important considerations.
Rather than restricting AI use, organisations are already beginning to reshape it by building oversight into how systems are used. That might mean making the AI’s working assumptions more visible and requiring humans to validate outputs. It might also require AI systems to signal their confidence in those outputs and to request manual checks. Over time, this reduces risk and creates an environment where people can work confidently with AI without losing sight of who is responsible.
This approach also challenges a common narrative. Much of the current discussion around AI focuses on job displacement, but the reality is more nuanced. The issue isn’t a simple replacement of human intuition and experience, but how those qualities evolve alongside increasingly capable systems.
Rather than removing the need for people, this managed shift reinforces it. Greater emphasis is placed on human-in-the-loop models, in which individuals with a deep understanding of AI can interrogate, challenge, and interpret system outputs.
A different starting point
So, what happens next? AI tool adoption, for sure. But it goes far deeper than that. Getting ready for AI natives means shifting to a different starting point – and learning to “live in the prompt”.
As AI natives begin entering the workforce and eventually move into leadership roles, the expectation won’t be that AI is introduced; it will be that it is already there. That shift reshapes how people think, learn, and approach tasks from the outset. It will also change how tasks are conceived, carried out and reviewed. The ability to configure, interrogate and challenge systems will become as important as the ability to interpret their outputs.
For organisations, the question is no longer whether to adopt AI, but whether their processes, governance structures and training pathways are ready for a workforce that already assumes it – and will expect to work that way from day one.
Adnan Patka, Enterprise Manager – Blockchain, AI and Web3 at OVHcloud, on why the convergence of blockchain and AI heralds a brighter future for both
SHARE THIS STORY
Anthropic’s Mythos team recently assessed various LLM technologies, scoring ChatGPT against Gemini, Grok, Claude and others. The metrics included trustworthiness, but also user deception, sycophancy, encouragement of user delusion and co-operation with human misuse.
Clearly, there’s no doubt that AI can be an enormously positive transformational force: the development of AlphaFold, whose founders won the Nobel Prize, showed us that very clearly. But when AI is being assessed for encouraging ‘user delusion’, it’s clear that it still has a PR problem.
In fact, AI has more than just a PR problem: research from McKinsey highlighted that although almost two thirds (64%) of decision-makers reported that AI was accelerating innovation within their organisation, less than half (39%) could actually report positive financial impact on the business as a result.
The AI Trust Gap
AI doesn’t exist in a vacuum, and the interplay of AI with other technologies is beginning to come into focus. We see that blockchain has great potential to complement AI, providing the transparency and trustworthiness that it needs. In a recent study of Web3 professionals, almost three-quarters (70%) said that blockchain had the potential to fill AI’s trust gaps.
But where else are these issues coming from?
It’s overly glib to say that people and businesses don’t always trust AI. What our research told us was that 32% of Web3 experts believe that privacy is AI’s main challenge. Historically, AI systems have been shown to be biased, and data has often been collected by AI systems for training or other uses – sometimes without user permission. One in seven (14%) of the study also reported that they believed AI needed more transparency to be trustworthy.
Blockchain’s Potential for Supporting AI
Blockchain is well-placed to help tackle these issues. Blockchain was originally created as a database with clear traceable links between items, without the need for a single governing authority.
For example, consider an AI system which optimises food supply chains, allowing a supermarket to maintain lower prices. In an environment where people may distrust AI, they may suspect that the system is simply buying lower-quality stock and shipping it more affordably, potentially meaning shorter shelf-lives and less tasty tomatoes! However, blockchain systems can help to track the provenance of these items, proving where the tomatoes were grown, when, by who, and the subsequent shipping and treatment of them, all in a publicly accessible ledger.
Blockchain systems also automatically check the validity of transactions and link them together in a way that makes fraud very difficult. Changing one transaction generally requires changing the entire ‘course of history’ or having significant control (specifically, controlling over half of the nodes) of an entire blockchain, which is usually unfeasible in most large, public blockchains today.
Delivering Benefits with Blockchain
There are also a number of cases where blockchain can bring functionality benefits to AI systems. For example, as we saw, privacy is a significant concern for AI users, but blockchain systems are able to authenticate securely without exposing user credentials, through self-sovereign identity. This is especially useful where AI systems need to interface with other tools that might not be as trustworthy as the core system.
Blockchain principles can also support data exchange in federated learning systems, where data is shared between different machine learning platforms. Blockchain can establish the validity of this data while still preserving its integrity and user privacy. This means that AI can be trained on data that is consistent and robust – without breaking trust.
There are a huge number of applications for this kind of technology, but unfortunately, it’s not all plain sailing.
Blockchain’s Challenges
The industry is undergoing a huge change today. Interest in stablecoins – a crypto asset which is attached to the value of a more stable asset like a traditional (fiat) currency or other physical commodity – has quadrupled since October 2025. Large, well-established organisations are considering what private and public blockchains can do for them, which could ultimately have a positive and stabilising impact on the regular bear and bull cycle that the crypto market experiences.
However, some more negative parts of blockchain’s legacy persist. Almost two thirds (61%) of our research sample said that concerns about blockchain’s credibility were having an impact on its integration into the enterprise ecosystem. This includes blockchain’s links with the more unsavoury side of cryptocurrency, including cybercriminal activity. Over half (54%) simply said that there wasn’t enough understanding about blockchain’s capabilities and what it could really do, holding IT decision-makers back from making the most of it.
At the same time, this isn’t a one-way street. AI’s capabilities – in particular, automated, agentic AI – can carry out blockchain activities without needing user input. For example, AI can issue transactions or analyse data and summarise it for monitoring systems. There is tremendous potential for AI to support many different parts of the blockchain ecosystem, summarising trends, automatically carrying out tasks, or executing smart contracts without human intervention.
A Mutually Beneficial Relationship
Blockchain and AI have enormous potential to support each other. There are some organisations already starting to integrate blockchain into AI and vice versa, but it will still be a number of years before the relationship is mature. That said, two-thirds of the Web3 experts we surveyed (67%) believed that it’d take roughly one to four years, with only eight percent saying that it might take five to six years.
Indeed, given the enormous benefits that blockchain can bring to AI, not to mention the rapidly increasing acceptance of blockchain technologies across the financial sector in particular, the convergence of these two technologies is almost inevitable. If we can push past the challenges, the future is bright for both blockchain and AI.
Mark Talbot, Director, CS AI Initiatives at Appian, reasons that as organisations grow more capable with AI, the challenge shifts from proving its value to expanding access to it
SHARE THIS STORY
Many organisations have long treated improvement as something that arrives as a top-down effort, not something built with the people doing the work. Specialists designed new processes, discussed them in formal forums, and introduced them through large change programmes that often felt detached from daily work. For most employees, ‘transformation’ meant being asked to follow new rules, rather than designing better ways of working.
AI is starting to reverse that pattern. Instead of concentrating control and decision rights in a small, central group, modern AI tools give more agency to the people closest to the work. They can see what is not working, imagine better approaches, and use AI to help redesign and improve the processes they rely on every day. This shift – which can be described as the democratisation of AI – changes who participates in improving the business. However, it is worth remembering that this shift only works at scale when AI is embedded within a platform that maintains governance, visibility and control.
Process Improvement in the Hands of Many
Until recently, fixing a broken process often meant filing tickets, waiting for a slot on an IT roadmap, or hoping that a specialist team would eventually address the issue. Creating applications, building automations or redesigning workflows were seen as highly technical tasks. For most employees, waste and inefficiency were things to work around, not things they had the tools or authority to change.
That obstacle is now deteriorating, as long as organizations don’t lose sight of the fact that governance remains essential, particularly in highly regulated environments
AI agents, generative AI and conversational interfaces allow people across the business to shape how work is structured. Within this model, someone in operations can describe an outcome in plain language and have an AI system propose and embed the steps within existing processes. Within a governed platform, non-technical users can adapt existing solutions and automate repetitive tasks without waiting months for central support. At the same time, process insights give developers visibility into what is being built, enabling them to refine, standardise and scale applications more quickly across the organisation.
Data is opening up as well. Data fabrics and related architectures connect scattered information sources into governed layers that a wider audience can access safely. Instead of waiting on static reports, people can access relevant, trusted data when they need it, and use AI to interpret and apply it to their decisions.
When process insight and data access reach this level, best practices move beyond documentation or occasional training. Tools and workflows embed them into daily work, improving performance across the organisation.
Scaling Improvement Across the Organisation
As more individuals understand how their work connects to broader outcomes, organisations unlock a powerful driver of change. Process improvement no longer depends only on a small group of specialists. Employees can recognise when processes are inefficient or risky and have the means to address them at scale, inside an AI platform.
By encoding domain knowledge into AI assistants and digital coworkers within an enterprise-grade AI platform, organisations can share expertise across roles and levels. These AI-powered helpers do not replace professional judgment. They strengthen it. They surface options, highlight inconsistencies and provide context, while humans make the final decision. Over time, each interaction becomes both a learning moment and a new piece of institutional knowledge that organisations can capture and reuse.
In this model, process improvement is no longer episodic or confined to formal transformation projects. It becomes part of everyday work, inside a platform with AI tools that provide real-time feedback and recommendations.
AI, Noise Reduction, and Better Oversight
This shift raises a key question: if AI platforms make analysis, decision support, and process design more accessible, what happens to deep expertise?
There is a concern that easy access to AI advice might weaken people’s understanding. If answers are always a prompt away, will teams still develop the knowledge that comes from working through complexity? If people follow AI suggestions without grasping the logic, how meaningful can human oversight really be?
Over-reliance on instant guidance can create only surface-level competence. People may treat AI outputs as instructions rather than as inputs to their own reasoning.
On the other hand, used well, AI can create more room for expertise, not less.
By handling repetitive tasks and routine decisions, AI reduces the volume of low-value work that consumes people’s time. Teams can then focus on exceptions and refine how they make decisions. Instead of dealing with every routine request themselves, they can focus on work where context and experience matter most.
When AI removes more of the routine burden, teams have more capacity to focus on judgement, process design and oversight. That helps build expertise while keeping improvement connected to the wider goals and governance of the business.
Shaping AI, Not Just Living With It
As organisations grow more capable with AI, the challenge shifts from proving its value to expanding access to it. AI is moving from something that happens to the workforce to AI being something that is built and refined with the workforce.
Organisations should treat people as partners in shaping AI, rather than as operators of automated systems. When AI platforms can be combined with process visibility and human judgement, employees can have an outsized effect on the systems around them. They can influence how work is structured and how decisions are made. In that sense, AI redistributes who participates in designing better ways of working, and creates an opportunity to anchor that shift in thoughtful design and human expertise.
Daniel Ehnhage, Head of AI Transformation at Unit4, on why those that put people and capability at the centre of their AI strategy will unlock far greater and more sustainable value than those led by technology alone.
SHARE THIS STORY
As the head of AI transformation, it might sound counterintuitive to suggest that artificial intelligence is not the most important part of my work. It makes a significant contribution to the radical change we are looking to achieve, but the technology itself is only about 10% of the solution. A significant part of the planning and investment must be based around addressing issues like the integration of siloed information systems, the building of the organisational capability required to adopt AI safely and finding the right business case. The key is to understand that adopting AI is not only about improving existing processes – it’s about gradually reshaping how we work in a sustainable way. The goal should be phased, practical improvements that build maturity over time.
This can be daunting for any organisation that has well-established operating practices. It requires a deliberate shift from problem‑solving to rethinking how value is created across the organisation. AI is far more capable and can empower your teams to find new solutions such as gathering more intelligence about market opportunities to improve productivity and decision making. The focus should be on enabling internal teams to work smarter through safe, responsible AI adoption. If your organisation is prepared to embark on such change, you must recognise AI becomes most powerful when you bring the right data together. Today, many organisations, including ours, are still maturing in this area. Successful adopters of AI prioritise building data readiness step by step so AI can create real value without overpromising.
Obviously, the role of AI transformation then becomes much broader with the added challenge of having to implement change without disrupting existing business performance. Consequently, there are some key areas where organisations must focus their attention, beyond ensuring they pick the right AI tool…
Structural Change – Put the AI Board in Place
AI transformation evolves how organisations work. It does not replace everything we humans do today. The goal is to focus on practical, high‑value use cases that improve productivity, quality, and employee experience without creating disruption. A widely debated expression of this change is the concern that AI will replace human employees. Personally, I think this lacks imagination around the positive impact that AI can have on a workplace. Yes, it may reduce the number of repetitive, mundane tasks, but more importantly it will create new ways of working and collaborating.
However, given how rapidly the technology is moving it is critical your organisation puts the right safeguards in place and agrees policies about ethical usage. That requires adoption of a cross-functional AI Board to provide a framework for embracing AI which will manage the impact of the structural change. This provides focus for your organisation’s approach to AI. The goal should be to agree which tools offer the most benefit for your teams and concentrate on exploiting use cases that will deliver the most benefit.
The AI Board should be responsible for establishing the governance structure to help the IT and cybersecurity teams to ensure the use of AI is not creating new vulnerabilities. It should provide clarity and safeguards so employees can use AI confidently and responsibly. Our goal is to enable safe experimentation – not restrict innovation.
People Change – Enabling Collaboration and Experimentation
The ambition should be to get employees excited about the potential of AI to open up new ways of working that can lead to rewarding opportunities and exciting new challenges. Indeed, it is widely accepted that helping your people to accommodate the change is the biggest challenge you will face, taking up about 70% of the time required to implement the technology. This is because successful implementations depend on collaboration between distinct teams, which in turn depends on breaking down barriers, both for individuals and teams.
For example, imagine being able to use AI to analyse data from diverse systems such as customer service, product development and marketing to identify new opportunities to support customers.
Integrating these data sources could be seen as interfering with distinct job functions, so for all employees it is critical to educate them on what will be expected of them and a good start point is explaining how they will be measured. It could include simple measures such as demonstrating usage of AI tools, but if an organisation wants employees to adopt the technology it is also important to empower them through training.
With the right support, employees will want to experiment, which should also enable them to understand use cases for AI in their work and the competencies they need to develop. This can be achieved through opportunities for cross-functional teams to explore new ways of working and innovating and should be encouraged by senior leaders. It is crucial they set the right tone, support initiatives, celebrate successes and listen to employee feedback.
Business Change – Building the Right Business Case
The business case is not just about saving money to solve a specific problem. It is too easy to look at the saved hours and productivity gains from adopting AI as the sum total of the investment costs you must deal with. There are a number of internally focused requirements that you must build into your thinking about AI transformation. There will be costs around the integration work to enable AI to access data from disparate systems. Competence development must be a top priority. Time must also be allocated to the process of change management and how it may disrupt existing business processes. Security must be a top consideration. These are all internally focused tasks, but you must look at them if you are to capitalise effectively on your AI investment.
It is tempting to become overly excited by the potential of AI as a technology, and certainly it will bring dramatic change to organisations in the years to come, but having experienced the factors necessary for successful transformations, it is absolutely critical senior leadership teams approach AI-enabled change with cool heads and clarity on what they want to achieve. Above all, they must remember success is not dependent on the implementation of the technology, but predominantly on bringing employees with them on the journey. Many commentators talk about the rise of AI-first organisations. Those that put people and capability at the centre of their AI strategy will unlock far greater and more sustainable value than those led by technology alone.
Our cover star Shadman Zafar, Founder & CEO of Vibrant Capital, is building a CIO-led model for enterprise transformation. Vibrant Capital is an operator-led investment and company-building platform focused on scaling AI in the real economy. “We don’t spray investments across hundreds of AI startups. We curate a portfolio with purpose – selecting companies that solve the real mission-critical problems CIOs face in scaling AI adoption.”
FNB: Redefining Data Science in Commercial Banking
We also hear from Yudhvir Seetharam, Chief Analytics Officer at South Africa’s First National Bank (FNB) on a data science journey characterised by curiosity, culture and the drive for a competitive edge. “Ours is a holistic approach focusing on the customer,” he explains. “Understanding the context of each customer journey and then using that context so that when we interact with you, we’re able to drive the right conversation with the right customer, at the right time, through the right channel and for the right reason. These ‘five rights’ make our interactions with clients more impactful.”
Virginia Farm Bureau: An Enterprise CIO’s Journey
Shifting focus to the world of insurance at the Virginia Farm Bureau, we spoke withan Enterprise CIO at a complex mission-driven organisation. As he approaches retirement, Patrick (Pat) Caine reflects on his career as a CIO and the centennial of an organisation renowned for resiliency, collaboration, commitment to a greater cause, diversity and service to its members. “In my role as CIO, I’ve always been that person who connects the dots between business needs and technology execution. Virginia Farm Bureau is digitally relevant, collaborative, and well‑positioned for the future.”
Mastercard: Protecting Trust in the Digital Economy
Michele Centemero, EVP Services at Mastercard Europe explains why promoting awareness, stronger collaboration and data-sharing, and continued innovation of payments ecosystems, will be critical in reducing the impact of scams and protecting trust in the digital economy. “The combination of AI, robust identity controls and open banking can help protect consumers from scams, whether across card and account‑to‑account payments or in fraudulent account openings.”
Thales on AI Security: How FinServ’s Budget Priorities Signal a Boardroom Shift
Todd Moore, Global VP – Data Security Products at Thales, reveals why making AI security a boardroom priority today, will help firms position themselves to capture competitive advantage, safeguard customer confidence, and define the future of secure innovation. “Balancing AI’s opportunity and risk means embedding security at every stage, from design to deployment and ongoing monitoring.”
Paymentology: The First Live AI-Agent Payment Is a Test for Credit Infrastructure
Thomas Benjaminsen Normann, Product Director at Paymentology, dissects the future for agentic payments and the progress still to be made. “Agentic payments demand something more granular: a clearer account of who or what acted, under what limits, and with what right to create a liability on the customer’s behalf.”
Also in this issue, we hear from Publicis Sapient, on why asset managers must redesign their enterprise for AI-driven decision intelligence; learn from Bitpace why the most resilient payments infrastructure will be the one with the most adaptability; rank the AI maturity of 12 of the largest payments networks in the latest Evident AI Index; and round up the key FinTech events and conferences across the globe.
Todd Moore, Global Vice President, Data Security Products at Thales, on why making AI security a boardroom priority today, will help firms position themselves to capture competitive advantage, safeguard customer confidence, and define the future of secure innovation
SHARE THIS STORY
Financial Services organisations are responsible for some of the biggest growth in the global economy. Equally, they’re some of the most vulnerable. Like many other sectors, they’re racing to embrace AI, but with adoption comes new security risks.
According to Thales’ Data Threat Report: Financial Services Edition 81% of FinServ organisations are now investing in GenAI-specific security tools, with nearly a quarter using newly allocated budget. This surge in funding marks a turning point: AI security has moved from being an IT concern to a boardroom priority.
The fact that new budget lines are being carved out specifically for AI security signals a fundamental shift in corporate strategy. Boards increasingly recognise that protecting AI systems is as critical as safeguarding payment rails or core banking infrastructure. For an industry built on trust, resilience, and regulatory compliance, this investment wave shows how central AI has become to both risk management and competitive growth.
Balancing AI Innovation and Security
While FinServ organisations are aware of the security risks AI poses, they’re also seizing upon the opportunities it presents. The report has found that in 2024, FinServ businesses outpaced the broader market in AI deployment, leading in enabling employees to use AI and ahead in AI integration, which has continued into 2025. Additionally, 45% say they’re in the ‘integration’ or ‘transformation’ phases of their GenAI journey, compared to just 33% across wider industries.
AI’s ability to accelerate services, automate processes, and analyse data at scale makes it an exciting prospect, especially in the financial sector. This makes securing AI systems a priority for FinServ organisations, with increased GenAI integration reflecting developing organisational maturity and progress beyond experimentation.
The Risk
Yet the scale of opportunity is matched by the scale of challenge. AI systems require vast amounts of structured and unstructured data to conduct analysis and make recommendations.
For FinServ organisations, this often includes highly sensitive customer and transactional information, proprietary algorithms, and records bound by strict regulatory oversight. The risk is not only about whether AI systems themselves are secure, but whether the data they’re working from is accurate, as well as whether their adoption inadvertently creates new routes to data exposure and exfiltration.
Businesses need a clear strategy to fully understand how AI models are operating within their IT infrastructure, the applications they’re interacting with, and the data they’re accessing and pulling from.
The Response
Balancing AI’s opportunity and risk means embedding security at every stage, from design to deployment and ongoing monitoring. Newly allocated budgets for AI security, with nearly a quarter of FinServ firms making such investments, show how central AI has become to board-level strategy. These investments move firms beyond reactive fixes to proactive frameworks that evolve with the technology. AI security is no longer just an IT concern, it’s a strategic priority requiring collaboration between security, compliance, and business leaders. By factoring risk into early planning, organisations can align innovation with responsibility and build resilience for the long term.
Pioneering AI Security
Building on investment in AI-specific security is only the beginning. As scrutiny intensifies, the firms that will lead are those that treat AI security as integral to business strategy, not a bolt-on layer. Success will require visibility into how models behave, continuous validation against emerging risks, and adaptive controls that evolve with the threat landscape.
The financial services organisations that embed these safeguards into their core infrastructure will protect sensitive data as well as setting a benchmark for resilience and trust in an AI-driven economy. By making AI security a boardroom priority today, these firms position themselves to capture competitive advantage, safeguard customer confidence, and define the future of secure innovation.
Thales: AI is the New Insider Threat
Thales 2026 Data Threat Report Finds 70% of Organisations Rank AI as Top Data Security Risk
Data security has taken centre stage as the success of enterprise AI initiatives increasingly hinges on consistent, controlled access to proprietary organisational data sources. The 2026 Thales Data Threat Reportexamines the complex calculus that organizations must undertake to enable innovation while securing their most valuable asset – their data.
This research was based on a global survey of 3,120 respondents fielded via web survey with targeted populations for each country, aimed at professionals in security and IT management.
Lee Fredricks, Director – Solutions Consulting, EMEA at PagerDuty, on why technology leaders should see 2026 as a time for operational resilience to shift from ambition to accountability
SHARE THIS STORY
Technology leaders should see 2026 as a time for operational resilience to shift from ambition to accountability. In 2025, too many cloud services outages and disruptions took place across the public and private sectors, and now regulatory, technological and cultural pressures are converging to say that enough is enough.
Outages often translate into broader repercussions for the organisation, including revenue impact, customer churn, share price pressure and potentially regulatory reporting obligations. Operational metrics must now be discussed alongside financial KPIs at the board level. C-suite leaders understand accountability, especially within the very regulated financial sector.
DORA’s First Birthday
It’s now been one year since the implementation of the Digital Operational Resilience Act, or DORA, introduced by the EU to strengthen the digital resilience of financial institutions. By now, organisations have had time to consider moving from mere compliance to creating a competitive edge from their investments.
Enterprise tech leaders are in the middle of a balancing act. They’re managing ongoing modernisation and transformation initiatives while navigating multi-jurisdictional regulatory scrutiny. At the same time, they face constant pressure from the board and must meet evolving customer needs—all competing for immediate attention. The stakes have never been higher. Operations teams are no longer viewed as a back-office IT function. Their success in keeping the organisation running and driving revenue is now a board-level concern.
For organisations today, IT is business delivery.
A year of DORA has seen organisations make the shift from focusing solely on mere compliance to setting meaningful demonstrable testing, third-party risk visibility and strictly mandated incident reporting timelines. Financial firms have lessened their exposure to risky situations. Payments providers aren’t only reliant on a single cloud region or SaaS supplier, or unable to provide evidence of real time incident response efforts and auditable logs after a disruption.
One benefit of these overall systemic improvements is enhanced supply chain accountability. Financial institutions and their technology partners are both liable for potential penalties and reputational risk, which makes it highly critical that they can prove their resilience capabilities.
Nevertheless, operational resilience is a continuous discipline. A fragmented incident response can expose firms to regulatory and reputational risk again and again if not addressed systemically. As such, many organisations are looking toward AI agents as part of a move towards ‘no-touch’ operations.
From Autonomy to Self-Healing
Under set policies, autonomous agents can handle incident response and operational tasks, such as detection, triage and remediation. AI agents deployed in operations may become the backbone of L1 (first contact) and L2 (more skilled) support. Contrast this with the traditional, reactive, ticket-driven model of IT. The industry can move much faster and with a higher successful close rate. Leveraging intelligent automation reduces mean time to detection/resolution and KPIs around lower incident volumes reaching L3. Additionally, it can lead to improved service availability percentages. Well integrated agents that actually support existing operations teams also help manage the issues around talent shortages faced by many organisations.
A typical incident lifecycle with agentic processes includes several stages depending on the model, but can be summarised as: Anomaly detected, correlated with recent deployment, a remediation script triggered and a human notified if set thresholds were breached. Such no-touch operations are golden in any sector, but particularly with industries such as digital banking and retail, where peak traffic periods demand near-instant response and poor customer experience is a powerful motivator for users to instantly change providers.
IT Standardisation
In addition, consider standardisation as part of strategic infrastructure best practices. There is a role for central operations clouds and operational ‘golden paths’ as solid foundations for reliable operational scale and dependability. Standardisation enables consistent, scalable operational excellence especially across large, distributed enterprises. ‘There is one way and it is the right way’ can be a great time and stress saver for operational teams – particularly if a regulatory notification and clear evidence is required.
For example, a global bank might define a single golden path for deploying customer-facing applications with pre-approved monitoring, incident response workflows, and regulatory reporting templates built in. In an outage, teams follow the same process and automatically capture the evidence required for regulators, avoiding confusion, delays, and compliance risk.
All of these possibilities take us to an exciting new place for an evolved set of developer and operational roles. When organisations enable AI to reshape daily engineering work away from manual firefighting and low-value work it frees headspace and time for developers and engineers to move into more architectural thinking and intelligent oversight of automated systems. These augmented teams will be empowered to manage simple situations instantly and devote more time and attention to the more difficult issues – the edge cases and the strategic necessities.
Enabling Agentic AI
Using another lens, businesses with agentic IT operations capabilities support their current talent, extending their reach and the speed of their response. The winning organisations will be those who deploy agents strategically, freeing up humans for that higher-value work – i.e. L3 expert support – and setting new standards for operational excellence that customers can rely on. Ideally this means making commensurate investment in existing people, training and organisational change management. A culture of continual upskilling and forecasting that points humans to where they make the best impact will be just as important as the autonomous tech tools working alongside them.
Autonomous agents allow many new services, and one of those can be described as self-healing operations. This evolution of the operations world is where predictive detection, automated remediation and embedded resilience all coalesce. With an autonomous process of testing, maintenance and remediation, organisations can focus on finely measuring improved customer trust. They can also enjoy the productivity and revenue benefits of high business continuity and availability.
AI is still a new technology, and many are legitimately concerned with the concept of autonomous agents. There is a need for clear guardrails, audit trails and explainability in automated remediation, and many technology partners have invested in their ability to support across these areas. Moreover, firms must maintain direction with policy-driven automation rather than uncontrolled autonomy, particularly in regulated industries.
Mandate Operational Excellence
This year is very likely to reward organisations that treat operational resilience as core to their business strategy. Those investing in automation, standardisation and governance will set the pace for their industries in an AI-enabled and increasingly autonomous world.
Regulators are already expanding their scrutiny and reliability expectations beyond financial services firms. Across the world, jurisdictions are increasingly looking to strengthen their economies and digital services in particular through resilience and cybersecurity measures. At the same time, agentic operations, and the organisational performance benefits they support, will rapidly become table stakes technology in all sectors. Inevitably, customers will judge brands on digital reliability as much as price or product features when evidence of outages are a click or a headline search away.
Start now. Audit internal incident response maturity, review the potentially complex web of third-party IT dependencies and identify where automation makes clear business sense. While resilience is an investment in compliance, it is also critical to ensure customer trust and future stability.
Jamil Jiva, Global Head of Asset Management at Linedata, on why the next chapter of AI-driven finance will be shaped not just by technology, but by creativity
SHARE THIS STORY
Beyond Data: Where AI Finds Unexpected Inspiration
The discussion about training AI largely focuses on concerns that accessible, human-generated data is limited and may soon run out completely. If this is the case, how can technology that depends on a seemingly endless stream of inputs to iterate, test, and adapt deliver the results we expect? AI relies on structured, high-quality data to thrive, but what happens when we run out of spreadsheets and financial models to train AI? We need new data sources to ensure it continues to learn, adapt, and deliver accurate insights. Video games stand out as offering some of the richest, most expansive, and complex environments for AI training.
At first glance, video games and financial operations seem to belong to entirely separate worlds. However, AI connects these domains, with models leveraging virtual-world training to tackle real-world financial tasks. Financial documents such as credit agreements and tax returns are often convoluted, unstructured, and labour-intensive to process. Therefore, AI designed to interpret such data must possess strategic reasoning, real-time adaptability, and advanced pattern recognition. So, could video games be the ideal training ground?
Contrary to popular belief, gameplay can significantly improve how people think, learn, and solve problems. The abilities required to excel at video games closely reflect the skills AI systems must acquire today.
Levelling Up: What Virtual Worlds Teach Machines
Practice leads to proficiency, a principle that applies to both humans and AI. Interestingly, many of the most significant advances in AI development have emerged not from conventional data training, but from taking creative approaches. Games push AI to emulate human thinking and sharpen its statistical intuition.
These game-trained models are neither expensive nor heavily reliant on resources, and they sidestep the issue of data scarcity. As a result, they are actively shaping the future of financial intelligence. The examples below offer a clear demonstration of the potential of gameplay.
Virtual Economies: Lessons from World of Warcraft
World of Warcraft, with millions of players interacting in an immersive and dynamic world, features an economy that closely mirrors real-world financial systems, complete with inflation, supply and demand cycles, and fraud risks. The game even inspired one of the most renowned epidemiological studies: when the in-game ‘Corrupted Blood’ plague spread unpredictably, scientists used it as a model for real-world pandemic simulations.
Financial models depend on vast, interconnected data networks, much like the economy in World of Warcraft. Organisations employ AI to continuously monitor patterns, detect anomalies such as fraud or misstatements, and optimise data extraction for financial reporting, mirroring the way AI analyses virtual economies.
Urban Chaos: GTA V and Real-World Simulation
While Grand Theft Auto (GTA) V is famous for its open-world chaos, researchers have leveraged its traffic systems and non-player character behaviours to train AI for applications such as self-driving cars, crime pattern recognition, and urban planning. At its heart, GTA provides a platform for AI to process vast amounts of unstructured data in real time.
Similarly, financial institutions manage millions of data points from a wide range of sources. Their AI tools must automatically extract insights, classify information, and normalise complex formats. GTA serves as a controlled yet intricate environment for simulating scenarios, enabling AI to optimise for real-world tasks through ongoing feedback loops.
Sandbox Creativity: Minecraft and Adaptive Thinking
Minecraft provides a sandbox environment where AI learns through exploration. OpenAI even trained an AI to play Minecraft by watching YouTube tutorials, closely mimicking the way humans learn. Similarly, any AI used by financial institutions must be able to self-learn from new document types and structures, adapting just as a Minecraft AI learns to survive.
Reinforcement learning, where AI improves based on feedback, is a key element of intelligent document processing. Thanks to its vast scalability and dynamic, hierarchical environments, Minecraft serves as an ideal setting for navigation and repeated feedback loops, helping models develop domain-flexible reasoning.
Multiplayer Mayhem: Dota 2 and the Art of Teamwork
Dota 2 stands out as one of the most complex competitive games ever created, presenting AI with challenges in real-time decision-making, strategic coordination, and adaptability. OpenAI Five, trained on the equivalent of 45,000 years of gameplay within just 10 months, managed to defeat renowned, professional human teams. As anyone who has mastered StarCraft knows, tactical adaptability is essential for gaining the upper hand.
Financial institutions operate in environments that are just as dynamic as the shifting levels of a video game. Market conditions, regulations, and data formats are in constant flux. AI must be able to adjust to new document structures, handle missing information, and navigate edge cases, much like AlphaStar adapts to an opponent’s unpredictable strategies.
From Pixels to Profits: Bringing Game Logic to Finance
Whether to streamline operations, mitigate risks, or make informed decisions in today’s data-intensive financial landscape, AI has the potential to fundamentally transform financial offerings, delivering personalised and evolving experiences that foster understanding and combine seamlessness with regulatory compliance.
Yet AI does not simply require more data from which to learn; it needs better data. Video games offer near limitless, pre-built, highly complex digital worlds where AI can test hypotheses, simulate scenarios, and refine decision-making models. By utilising these unique environments, AI is challenged to enhance its speed, accuracy, and efficiency.
The world of video games has many lessons we can learn when building AI, and given AI’s remarkable ability for transferable learning, it makes sense to leverage these pre-trained models to power essential financial workflows. It is more than just document processing; it is thinking, and the same intelligence that enables AI to defeat world champions in Dota 2 is now driving the next generation of financial AI solutions.
The next chapter of AI-driven finance will be shaped not just by technology, but by creativity. By embracing unconventional data sources such as the immersive complexity of video games, industry leaders will unlock new possibilities for personalisation, security, and customer engagement.
Richard Doherty, Head of Wealth & Asset Management, Publicis Sapient, on how asset managers must redesign their enterprise for AI-driven decision intelligence
SHARE THIS STORY
The asset management industry is entering a structural inflexion point. The first wave of AI focused on improving productivity through copilots and automation. The next wave will fundamentally reshape how decisions are made, executed, and governed across the enterprise. This is not a technology upgrade. It is an operating model shift.
Despite significant investment, many firms remain trapped in fragmented AI experimentation. A majority are yet to realise meaningful economic returns from AI, not due to lack of capability, but due to a failure to redesign how intelligence is applied across the organisation. The gap between ambition and outcome is not a technology problem. It is a structural one.
From Automation to Decision Intelligence
The industry conversation has evolved. The question is no longer whether to adopt AI, but how to scale it across the enterprise. However, most firms are still approaching this challenge through the lens of automation, identifying tasks that can be executed faster or at lower cost. This delivers incremental value, but does not address the underlying constraint: the structure of decision-making within the organisation.
Traditional operating models are built around sequential workflows. Work moves from function to function: research, compliance, operations, and distribution, each dependent on the previous stage. This creates latency, duplication, and fragmentation. Agentic operating models shift the focus from tasks to decisions.
Instead of asking “Which processes can we automate?”, leading firms are asking: “Which decisions can be augmented or owned by intelligent systems?”
This shift enables organisations to move from sequential workflows to parallel decision systems; from human-led analysis to AI-assisted reasoning; from periodic insight to continuous intelligence. The result is not a marginal improvement. It is a step-change in how the enterprise operates.
The Pressures Driving Change
This transformation is not happening in a vacuum. Asset managers face mounting structural pressures: margin compression driven by fee pressure and passive competition; rising operational complexity from regulation and product proliferation; and advisor capacity constraints that limit scalable growth. Agentic operating models directly address all three.
By automating complex workflows, rather than individual tasks, firms can significantly increase advisor and analyst capacity without proportional cost increases. Parallel decision systems reduce the time required to launch products, respond to market events, and deliver client insights. This compresses cycles from months to days. Continuous monitoring of guidelines, portfolios, and operational processes reduces exposure to regulatory breaches and operational failures.
These are not theoretical benefits. They represent measurable improvements in cost-to-serve, time-to-market, and operational resilience.
Not all Intelligence is the Same
To scale AI effectively, organisations must recognise that not all problems require the same type of intelligence. Enterprise AI operates across three distinct layers, and conflating them is one of the primary reasons AI initiatives fail to scale.
Deterministic systems execute predefined rules with complete consistency. They are essential for functions where there is zero tolerance for error, trade validation, settlement processing, and regulatory reporting. If a business outcome must be identical every time, deterministic logic remains the correct approach.
Predictive systems use historical data to forecast outcomes. Applied in areas such as portfolio risk modelling, fraud detection, and client churn prediction, they generate probabilities and insights, but they do not interpret context or make decisions independently.
Agentic systems operate where problems require interpretation, judgment, and contextual understanding, investment guideline interpretation, regulatory document analysis, portfolio insights, and client communication. These systems can reason across complex information, generate insights, and take action within defined boundaries.
The ‘Different but Valid’ Dilemma
A critical challenge in adopting agentic systems is understanding how they behave. Traditional software produces identical outputs. Agentic systems produce reasoned outputs.
This introduces what I call the ‘different but valid’ dilemma. An agent may take a different reasoning path from a human and arrive at a different, but still correct, conclusion. This variability is not an error. It is inherent to reasoning systems.
The real risk lies in hallucination, outputs that are not grounded in data or evidence. Managing this requires organisations to clearly define where variability is acceptable. All AI-driven processes sit on a spectrum: deterministic actions with no variability (trade execution), predictive actions with controlled variability (risk scoring), and agentic actions with higher variability (investment insights).
Leading firms design systems where agents perform reasoning, deterministic systems enforce execution, and humans retain oversight on high-consequence decisions. This balance enables both flexibility and control.
The Operating Model Shift
The most significant change is not technological; it is organisational. Traditional models are built on functional workflows. Agentic models are built on coordinated decision systems.
Consider what launching a new investment product looks like under each model. In a traditional model, it involves sequential handoffs between teams, compliance reviews the guidelines, operations configures the systems, and distribution drafts the client narrative. Each stage waits for the last.
In an agentic model, intelligent systems operate in parallel: compliance agents interpret guidelines, operations agents configure constraints, distribution agents generate client narratives, and governance agents validate outputs. This orchestration compresses timelines, reduces friction, and enables continuous decision-making. It represents a fundamental redesign of how work is performed.
Governance: the Foundation for Trust
Trust is the prerequisite for scaling AI. Without it, adoption stalls, not because the technology fails, but because the organisation cannot adequately explain or defend the decisions it makes.
Leading firms implement governance models built on three principles. First, explainability: every decision must be traceable and auditable. Second, authority boundaries: agents operate within clearly defined limits. Third, human oversight: high-consequence decisions remain under human control.
Regulatory expectations will continue to evolve, but one principle remains constant: organisations must be able to explain how decisions are made.
Scaling AI is a Leadership Challenge
Executives must take a deliberate approach across four areas:
Define the intelligence model: map business problems to deterministic, predictive, or agentic systems.
Build the foundation: invest in data, infrastructure, and orchestration capabilities.
Redesign the operating model: shift from workflows to decision systems.
Implement governance to ensure transparency, control, and compliance.
Start with high-value use cases and expand rapidly across the enterprise. The firms that act now will establish a structural advantage in cost, speed, and decision quality. Those that do not risk being constrained by legacy operating models that cannot scale with the demands of modern markets.
The Question is not if, it is Who
The industry is not simply adopting new technology. It is redefining how decisions are made. The firms that succeed will not be those that deploy AI tools in isolation. They will be those who design the right form of intelligence for each problem, redesign their operating models around intelligent systems, and scale agentic capabilities across the enterprise.
This shift is already underway. The question is no longer whether it will happen. The question is which firms will lead, and which will be forced to follow.
Martijn Gribnauis, Chief Customer Success Officer at Quant, on why Agentic AI will redefine financial services
SHARE THIS STORY
A recent Google Cloud survey showed that only 13% of finance organisations are currently using agentic artificial intelligence. This number needs to, and will rise when you consider that 88% of financial leaders are seeing ROI from generative AI already. Agentic is the next and most advanced evolution of artificial intelligence the world has ever seen.
Agentic AI is not on the way. It is here and already reshaping how forward-leaning financial institutions operate. In 2026, for IT and finance leaders to build an insurmountable competitive lead they must deploy agentic AI in every area where it can safely and effectively create value. The institutions that hesitate will find their business models under threat from familiar competitors and newcomers alike.
Reinvention of Core Processes
Agentic AI is poised to reinvent core financial processes. Bookkeeping, record maintenance, and period-end close are nearing complete automation. Month-end processes that once required late-night, stress-filled marathons will evolve into continuous, largely automated cycles. IT teams will no longer spend evenings on high alert waiting for failures.
This shift also frees IT leaders, finance teams, and operations functions from monotonous repetitive tasks. Instead of focusing on system uptime and manual reconciliation, they will collaborate with the C-suite on strategic initiatives that drive growth and revenue.
Understanding Why Adoption Is So Low
Despite the promise of Agentic AI, there is understandable caution. Some 80% of organisations have reported ‘risky behaviour’ from AI agents, and in the world of finance that is an alarming number. Finance is one of the most regulated, risk-averse sectors in the world. The fear of losing control remains the primary reason so few in the industry have embraced Agentic AI.
Loss of control and fear of catastrophic error
Financial leaders fear that an autonomous system could go ‘off script’, mis-route payments, misinterpret rules, or inadvertently cause compliance breaches. In finance, even small errors can trigger major financial or regulatory consequences.
Security and data privacy concerns
Large AI models require huge quantities of sensitive data. Organisations worry about breaches, cyber-attacks, or manipulation. An AI agent with improperly configured permissions could, in theory, execute fraudulent transactions or expose confidential customer information.
Bias and fairness risks
If AI agents make decisions using incomplete or fragmented data, they risk perpetuating or amplifying bias. At scale, biased decision-making can undermine customer trust and expose firms to legal and regulatory challenges.
Regulatory ambiguity and audit difficulty
Regulators are still determining how to govern agentic AI. Some organisations fear that early adoption could unintentionally violate rules or create future audit vulnerabilities.
These fears are legitimate, but not insurmountable.
Tackling the Adoption Barriers: A Practical Blueprint for Finance Leaders
To capitalise on Agentic AI’s immense potential, leaders must take a structured approach grounded in business value, security, and trust.
1. Start With Clear, Measurable ROI and Efficiency Gains
In finance, adoption accelerates when decision-makers see proof of value.
Start by automating repetitive processes. Agentic AI can handle tasks like data entry, reconciliation, invoice matching, and initial fraud checks faster and more accurately than humans. This leads to reduced operational overhead as automation lowers labour costs, shortens processing times, and reduces error rates. Demonstrating these savings through case studies or internal pilots is critical to changing minds.
AI agents can enable revenue growth by analysing huge data sets to identify new investment opportunities, optimise trading strategies, and generate personalised product recommendations. Each of these capabilities directly impacts top-line growth.
2. Strengthen Risk Management and Compliance Through AI
Agentic AI will improve risk management when deployed responsibly. This starts with real-time fraud detection. AI agents can monitor transactions continuously, identifying patterns that suggest fraud long before traditional systems would detect an anomaly.
Continuous monitoring is also incredibly helpful when it comes to compliance. AI agents excel at ensuring adherence to KYC and AML regulations. They can automatically maintain audit trails, identify missing documentation, flag anomalies, and escalate issues instantly.
Enhanced stress testing and scenario modelling can both be completed via Agentic AI. It can simulate complex market environments more dynamically than legacy tools, providing deeper insights into vulnerabilities and improving resilience. When showcased and presented in this context, agentic AI becomes a risk-reduction tool in the eyes of decision makers.
3. Directly Address Security and Trust Concerns
Trust is the cornerstone of adoption. Implement enterprise-grade security architecture that includes encryption, secure APIs, strict access controls, and continuous monitoring of agent behaviour. And, use explainable and transparent AI systems (XAI) so your finance teams understand the reasoning behind decisions. XAI helps provide interpretable outputs that support auditability and regulatory compliance.
Start small with a controlled, low-risk pilot. A proof-of-concept in a non-critical workflow helps teams understand the technology, gather evidence, and build internal support before scaling. Produce numbers based reporting that speaks the language of the people who make the decisions. Show, don’t just tell them how agentic will move the business forward.
4. Highlight the Competitive Advantage
Agentic AI adoption is not just an efficiency upgrade. It is a competitive imperative. AI agents create faster innovation cycles by accelerating product development, service delivery, and operational improvements.
They also provide superior customer experience. From instant account servicing to personalised financial recommendations, Agentic AI delivers the speed, personalisation, and convenience customers expect. Plus, it scales exponentially. No matter how many people call in at the same time, an agentic agent will answer immediately. Agentic AI reduces up to 86% of time spent in complex workflows that were traditionally handled only by people. This will be huge in getting ahead of your competition.
5. Build Momentum Through Internal Champions
Adoption increases when respected leaders advocate from within. Mid-level managers, AI-literate staff, or members of the C-suite who understand the technology can serve as champions. Use them and their beliefs to drive alignment, communicate benefits, and counter misconceptions. The more people from different departments and levels of the organisation that talk up the technology, the more likely you are to get buy-in.
Your Time is Now
Agentic AI will redefine financial services. The organisations that act today will build capabilities, insights, and competitive advantages that late adopters will not be able to replicate. Finance leaders must begin asking where agentic AI can support their business, where it can remove friction, where it can unlock growth, and where it can transform operations. The firms that act now will lead the industry. Those that hesitate will not get the chance to catch up.
The only remaining question for finance organisations is not whether agentic AI will change the industry, but how quickly they choose to deploy it.
Dr. Yvonne Bernard, CTO at Hornetsecurity, on meeting the challenge of managing the speed of AI adoption and harnessing its defensive capabilities while mitigating the risk of uncontrolled adoption
SHARE THIS STORY
The past year has been defined by acceleration. Threat actors rapidly embraced automation, AI, and social engineering. Scaling their tactics at unprecedented speed, while defenders raced to keep pace. Historically, defensive resilience evolves in step with attacker innovation, but in 2025 that balance began to falter.
In an analysis of over 6 billion monthly emails, Hornetsecurity’s Security Labs found that the volume of sophisticated threats grew faster than most security teams could adapt to. Malware-infected emails soared by 131%, scams increased by nearly 35%, and phishing attempts – powered by access to advanced AI – rose by 21% from the previous year.
Typically, attacks, even at volume, are easily filtered by good firewalls and secure email gateways. But the sophistication and AI-led nature of 2025’s boom made it even harder for organisations to defend themselves. The question now is: can security teams and businesses wrestle back control?
Evolving Cyberattack Landscape
AI enhances efficiency and precision. As such, cybercriminals use it to launch faster, more convincing and adaptive attacks, ranging from deepfakes to credential stuffing. As an example, there is a concerning trend of attackers increasingly using ‘MFA bypass kits’ to create deceptive login pages. These pages capture not only the user’s credentials but also have logic built in to handle MFA prompts as well. The unsuspecting user is then passed to the real login page for the target service and meanwhile the ‘kit’ grabs a copy of the user’s session token. This allows the attacker to impersonate the person and access their data.
Examples of such kits include Evilginx (open source) and the W3LL panel. Protecting against these attacks can be challenging, as they are adept at bypassing MFA safeguards. Threat actors often use compromised LinkedIn accounts, for example, to gain access to substantial information and connections. This enables them to impersonate trusted business connections. Paired with the weaponisation of Agentic AI, this will magnify existing vulnerabilities within an organisation, while introducing new ones that defy traditional containment models.
As it stands, the lack of oversight within organisations on the extent of AI’s adoption by cybercriminals has enabled the emergence of ‘Ransomware 3.0.’ Ransomware has evolved past simple encryption and exfiltration, with this next phase focusing on LLM-driven orchestration and a shift to data integrity manipulation.
To counter AI-accelerated compromises and ‘Ransomware 3.0’ in 2026, organisations must adopt a Zero Trust-based cyber resiliency strategy. This requires businesses to implement strong, non-phishable machine authentication, strict least-privilege access, and constant monitoring to protect the integrity of the data that users and AI agents can access. It should become the baseline expectations rather than aspirational goals for this year.
The Secret Value of ‘Least Privilege’ Access
Another strategy to proactively improve cybersecurity defences in 2026 is to enforce the principle of ‘least privilege’ access. This tactic grants users access only to the data that’s needed for their role. Limiting excessive access is important for preventing the potential for widespread data exposure and damage in the case of an account compromise.
Businesses, however, must strike a balance over access; if it’s too strict, it can hinder productivity and lead to shadow IT issues. Getting this balance right when it comes to privileged access is where sophisticated permission managers are invaluable tools to work with. They streamline the process and remove the guessing game of who and what to grant access to, thereby ensuring, in the case of an attack, that the entire organisation won’t be brought to its knees.
How CISOs are Adopting ‘Resilience, not Perfection’
The rate at which AI is advancing means not every organisation will be equipped with the tools or the know-how to tackle every AI-inspired attack. But as the saying goes, ‘prevention is better than cure’. It’s better to create a strong security culture than to continually chase after the next best tool.
Organisations can’t strengthen their resilience without involving every single person under their umbrella. That’s why CISOs must continue to invest in cybersecurity awareness programs.
These should include simulated AI-phishing attacks (phishing remains the number one attack vector) to test users and enable them to apply learnings from the modules.
If any user clicks on a phishing email, they should receive additional training at that very moment, to cement the learning. Over time, a good training system should automatically identify users who rarely fall for such attacks and reduce the training they receive while making the simulations they do receive more difficult. Conversely, giving persistent offenders additional bite-sized training and simulations can help improve security outcomes over time.
The key challenge for 2026 is managing the speed of AI adoption and harnessing its defensive capabilities while mitigating the risk of uncontrolled adoption. But with excellent training, cyberattack practice runs, and the adoption of Zero Trust principles, organisations will find themselves in a strong position.
About Dr. Yvonne Bernard
Dr. Yvonne Bernard is the CTO of Hornetsecurity by Proofpoint, Proofpoint’s business unit leveraging the Hornetsecurity product suite dedicated to managed service providers (MSPs) and small to mid-sized businesses (SMBs), providing next-generation cloud-based security, compliance, backup, and security awareness solutions that help companies and organisations of all sizes around the world.
Dr Megha Kumar, Chief Product Officer and Head of Geopolitical Risk at CyXcel, on whether our risk and regulatory frameworks and institutional cultures can keep pace with Agentic AI
SHARE THIS STORY
Within the next couple of years, Agentic AI is likely to progress from early stages of operation to be fully embedded within systems. Its expansion will be subtle rather than spectacular. It will integrate steadily into enterprise platforms, logistics networks, compliance workflows, cybersecurity operations centres and executive decision-support tools. Processes will move faster, operating expenses will decline and performance indicators will trend upward.
Yet these visible improvements mask a deeper challenge. The regulatory exposure, data governance pressures and erosion-of-trust risks associated with Agentic AI are being misjudged.
Unlike earlier AI applications designed primarily to generate outputs – whether text, imagery, or predictive insights – agentic systems are built to act. They sequence decisions, draw from multiple data environments, initiate consequential processes and function at scale with differing levels of human supervision. In sandbox environments this can seem contained and controllable. Over extended periods in live environments, however, sustained oversight, traceability and effective governance become significantly more complex.
Evolving Operational Complexity
There are two key challenges that businesses must address.
First, how do organisations monitor what agentic systems are doing once deployed? These systems evolve through updates, integrations and retraining and they interact with new data environments.
Second, how do you ensure responsible behaviour throughout the lifecycle? Regulators, policymakers and customers will likely expect firms to shift from compliance assurance to risk assurance and demonstrable evidence of trust and transparency.
The prevailing assumption is that human oversight will mitigate these risks. Human in the loop or human over the loop has become the default reassurance. In practice, however, that assumption breaks down far faster than many anticipate.
When a system works 95 per cent of the time, human reviewers limit their scrutiny. Behavioural science tells us that automation bias and complacency occur when automated systems are high-performing. Employees often become validators of AI outputs rather than critical examiners. The diligence gap widens gradually and then suddenly.
Facing Up to Difficult Questions
How do you incentivise employees to remain diligent checkers when the system mostly ‘works’? And how much time does effective oversight actually require? True review is not a cursory glance at a dashboard. It involves interrogating assumptions, validating inputs, checking context and assessing downstream consequences. In many cases, meaningful oversight may take nearly as long as performing the original task manually. When checking becomes more costly than doing the job yourself, pressure to ‘trust the system’ intensifies.
And what happens to accountability when oversight exists on paper but not in practice? Governance documentation may show layered review structures, escalation pathways and audit processes. Yet if humans are functionally disengaged, responsibility becomes dispersed. When errors surface, organisations may struggle to attribute fault – was it the model design, the data, the integrator, the operator or the reviewer who signed off without fully scrutinising?
Regulators are only beginning to grapple with these realities. In jurisdictions such as the European Union, the EU AI Act introduces risk-based obligations, documentation requirements and human oversight provisions. These are important steps, however, the operationalisation of those requirements in dynamic, agentic environments remain untested at scale. Compliance on paper will not automatically translate into resilient governance in practice.
Addressing the Trust Challenge
Beyond regulatory exposure, there is a broader trust challenge emerging.
As Agentic AI systems scale across industries, they will generate vast volumes of automated outputs – reports, communications, risk assessments, content, decisions and transactions. If errors or manipulations spread through interconnected systems, confidence in digital outputs may erode.
In geopolitically sensitive contexts, this has profound implications. Agentic systems interacting with external data sources could amplify disinformation, introduce biased datasets or make decisions based on manipulated inputs. The speed of automation may outpace the speed of verification. Trust, once diluted, is difficult to restore.
Data protection risks will also intensify. Agentic systems frequently require broad access privileges to perform tasks effectively. They may access internal databases and personal data and interact with third-party platforms. Each interaction creates potential exposure points. A single misconfiguration or prompt injection attack could trigger cascading consequences across systems.
The next phase of AI adoption will not simply amplify productivity: it will amplify regulatory, legal and reputational risk. This moment therefore demands serious scrutiny before agentic AI becomes deeply embedded in business infrastructure.
The Moment for Action has Arrived
So, what should organisations be doing now?
To begin with, organisations need to look past superficial, tick-box compliance. Effective governance cannot live solely in policy documents – it must function in day-to-day operations. This means investing in continuous monitoring capabilities, robust audit trails and real-time anomaly detection tailored specifically to Agentic AI behaviours.
In parallel, incentive structures should be redesigned. Meaningful human oversight will not happen if it is treated as secondary to speed or output. If employees are expected to provide meaningful review, organisations must allocate time, training and authority accordingly. Performance metrics should reflect risk management responsibilities, not just output rate.
Clear lines of accountability are equally important. Senior leadership and boards should determine who carries ultimate responsibility for outcomes produced by agents. Where third-party vendors are involved, responsibilities must be contractually and operationally defined. Incident response mechanisms should be rehearsed in advance, rather than presumed to work when pressure is high.
Expertise must also be integrated across functions. Legal, risk, compliance, cybersecurity, data protection and operational teams should be engaged from the outset. Deploying Agentic AI is not simply a technical upgrade – it reshapes the organisation’s risk profile.
Finally, resilience demands deliberate stress-testing. Leaders should examine not only pathways to success but how models fail at scale. How would the organisation respond if a system update embedded systemic bias, if an integration vulnerability enabled unauthorised activity or if automated actions eroded customer confidence? Rigorous scenario exercises, however uncomfortable, are essential to building genuine preparedness.
As Agentic AI advances, Risk Management Should Match its Pace
None of this is an argument against adoption. Agentic AI presents meaningful productivity improvements and the potential for sustained competitive differentiation. Organisations that deploy it with discipline and foresight may secure a measurable advantage. The danger lies not in adoption itself, but in pursuing acceleration without knowing the risks and putting the right guardrails in place.
The coming two years are critical for businesses. Before these systems become deeply embedded in core processes, organisations have an opportunity to shape the control environment around them. However, once agentic systems are fully embedded, retrofitting controls will be far more difficult and costly. Leaders must therefore treat this period as a design phase for oversight, not merely a race for competitive advantage.
Agentic AI is advancing rapidly. The defining question is whether our risk and regulatory frameworks and institutional cultures can evolve just as quickly.
Adam Spearing, VP of AI GTM EMEA at ServiceNow, on why those that invest in AI foundations now will shape their operating models on their own terms
SHARE THIS STORY
Much of the debate around AI still centres on pilots: which tools to test, which use cases to prioritise, which risks to manage. Executive teams commission proofs of concept, establish governance forums and assess compliance exposure. Far less scrutiny is applied to the consequences of waiting.
Traditional technical debt is familiar territory for CIOs. It stems from shortcuts, ageing platforms and deferred upgrades. It builds over time and is eventually addressed through structured modernisation programmes. Visible in legacy code, brittle integrations and manual workarounds. It appears on risk registers and capital plans. Leaders know how to describe it and, in principle, how to resolve it.
Forward-looking technical debt is different. It arises when organisations postpone the foundational changes needed for new ways of working. It is not created by past expediency, but by present hesitation. And it accumulates faster.
AI Adoption
In the context of AI, the effects are already emerging. Each quarter spent debating readiness instead of building it increases the distance between legacy operating models and AI-enabled competitors. As models improve and user expectations shift, that distance widens, reshaping competitive baselines. What begins as a modest capability gap can harden into structural disadvantage.
While companies debate whether to adopt AI, the margin for strategic choice narrows. Many organisations frame AI adoption as a binary decision: adopt now or wait until the technology matures further. In practice, the room for discretion is smaller than it appears. Time spent stalled in pilots or governance loops increases the gap between internal capability and market expectation.
More than 75% of organisations are expected to face moderate to severe AI-related technical debt in 2026, predicts Forrester. The issue will not simply be missed efficiency gains. It will be structural misalignment between how their systems operate and how work is increasingly done.
This misalignment often appears gradually. Teams rely on manual data preparation because underlying systems cannot support automation. AI tools are layered onto fragmented architectures and deliver inconsistent outputs. Employees experiment with external tools because internal platforms cannot provide the functionality they need. Each workaround creates further fragmentation.
Over time, these patterns compound. Integration backlogs expand. Security and risk teams struggle to enforce consistent controls across proliferating tools. Data governance becomes reactive rather than designed. What began as caution begins to constrain strategic options.
The AI Paradox
Here’s the paradox: organisations are either rushing into unsuccessful AI pilots that create immediate technical debt, or they’re avoiding AI entirely and creating forward-looking debt through inaction. Both paths lead to the same place – systems that can’t support the future of work.
AI isn’t just another technology layer to bolt onto existing infrastructure. It’s fundamentally changing how people interact with systems and how work gets done. Increasingly, AI becomes an interface through which employees access information, execute tasks and navigate processes. When AI becomes the interface – not just for customers but for employees navigating their daily tasks – organisations without AI-ready foundations will find themselves unable to compete on speed, efficiency, or experience.
The companies that hesitate aren’t just missing out on automation benefits today. They’re building a deficit that grows exponentially as AI capabilities advance. Each new model release, each competitor’s successful implementation, each customer expectation shift adds to the debt. Each significant model improvement raises the performance benchmark across the market. Unlike legacy systems that degrade slowly, this gap accelerates.
From Avoidance to Advantage
Breaking free from forward-looking technical debt requires a fundamental mindset shift. This isn’t about buying more technology or launching more AI pilots. It’s about creating the conditions for sustainable AI adoption that builds capability rather than complexity.
The organisations succeeding with AI aren’t the ones with the biggest budgets or the most aggressive rollouts. They’re the ones that took a deliberate, phased approach to ensuring their data, systems, and culture could support AI at scale. They treated readiness as an operational discipline rather than an innovation side project. They understood that AI adoption isn’t a destination, it’s a continuous capability that requires solid foundations.
This starts with honest visibility into current technology estates. Leaders must understand what systems can realistically support AI workloads, where data quality creates barriers, and which processes are ready for automation. Only then can organisations introduce AI incrementally, modernising systems where necessary rather than forcing new capabilities onto brittle foundations. Without that clarity, AI risks being layered onto structural weaknesses.
Modernisation therefore becomes targeted. Consolidating fragmented workflows, standardising data models and reducing unnecessary integration points increase the feasibility of scaling AI across multiple use cases. Early deployments focused on well-defined processes with clear data lineage can build internal confidence while strengthening governance practices.
Clear Debt to Stay Competitive
Forward-looking technical debt does not appear on a balance sheet. It shows up in slower product cycles, manual workarounds, integration backlogs and frustrated employees. It surfaces when competitors deliver AI-assisted services as standard and customers begin to expect the same everywhere. By the time these symptoms are visible, the underlying gap has already widened.
Timing therefore becomes a strategic variable. AI capability builds cumulatively: early investment in clean data, modern workflows and interoperable systems creates a base for continuous improvement. Each iteration becomes easier, faster and more reliable. Those that delay face the opposite trajectory: increasing complexity, rising retrofit costs and shrinking room for strategic choice.
The real issue is not adoption in principle. It is whether leadership teams are prepared to treat readiness as urgent rather than optional.
Reducing forward-looking technical debt requires acting before competitive pressure dictates terms, aligning technology modernisation with operating model reform, and accepting that disciplined progress now is less risky than accelerated catch-up later.
AI adoption will continue irrespective of individual organisational hesitation. Vendors will continue to refine their offerings. Regulators will clarify expectations. Customers and employees will adjust their behaviours. Those that invest in foundations now will shape their operating models on their own terms. Those that delay risk reacting to a competitive gap that is already commercially significant.
Adonis Celestine, Senior Director – Global Automation Practice Lead at Applause, on the rise of AI and why In a world of autonomous systems, trust is the ultimate competitive advantage
SHARE THIS STORY
Every generation of technology has its defining disruptor – the force that rises above the rest and reshapes its environment. In the mid-2000s, Marc Andreessen captured the moment when digital systems began transforming entire industries with his famous line: “software is eating the world”. At the time, software was the apex predator of technology, defining how value was created and delivered. Today, that hierarchy has shifted. Artificial Intelligence (AI) has reached the top of the technology food chain. Not just accelerating software, but fundamentally reimagining how it’s created, tested, and deployed.
AI is no longer just a tool; it is a co-creator. Developers now rely on AI daily to translate high-level intentions into working code. A practice sometimes known as ‘vibe coding’. Tasks that once took months can now be delivered in weeks, days, or even minutes. The pace is exhilarating, but it introduces challenges that traditional quality assurance (QA) practices were never designed to meet. And if QA cannot keep up, speed will come at the cost of reliability and trust.
When AI Outpaces QA
Conventional QA depends on predictability. Features are defined, code is written, and test cases verify the expected behaviour. However, AI disrupts this traditional model. Generative and Agentic AI systems don’t simply follow instructions; they interpret them. These systems adapt to context, learn from data, and can produce different outputs from the same prompt, influenced by factors such as training, temperature settings, and the model’s probabilistic nature. With development cycles now measured in minutes, traditional QA handoffs are often impossible.
This has led to a growing gap between speed and certainty. Teams can ship products faster than ever, yet it’s becoming much more difficult to ensure consistent, ethical, or safe behaviour in real-world conditions. Enterprises are already experiencing AI-powered features that fail in ways conventional testing could not anticipate, undermining trust and creating new risks.
Hidden Risks in Autonomous AI Workflows
AI-driven development introduces blind spots that traditional QA often struggles to detect. One key issue is context drift. This occurs when AI performs well in controlled testing environments but behaves unpredictably when faced with edge cases, cultural differences, or ambiguous inputs. For example, a customer-facing chatbot might pass functional tests but produce biased or misleading responses when deployed on a global scale.
Another challenge is compound autonomy. When multiple AI agents are involved in code generation, testing, and deployment, the system may begin to validate its own processes. Without human oversight, errors can propagate unnoticed. An AI agent might ‘approve’ certain behaviours because they statistically align with previous outputs. Rather than meeting user or business expectations.
Invisible change also complicates QA efforts. AI models continuously evolve through processes like retraining, prompt tuning, or data updates. A feature that worked flawlessly last week may function differently today. Traditional regression testing often fails to capture these subtle but significant shifts.
Most critically, AI workflows blur the lines of accountability. When failures occur, it can be unclear whether the issue lies with the model, the data, the prompt, the integration, or the deployment pipeline. QA teams must continuously validate not only the outputs but also the decision-making processes behind them.
Redefining Quality and Trust in an AI World
Slowing AI development is neither practical nor beneficial. Organisations must redefine quality in a probabilistic, AI-driven environment. Quality now extends beyond just correctness. It involves ensuring that systems operate reliably in real-world scenarios. This shift requires moving from static test cases to continuous, adaptive validation.
QA teams must evolve into ‘quality intelligence’ teams, broadening their responsibilities from simply detecting defects to actively fostering trust in AI systems. AI-assisted testing is crucial in this process. It can automatically generate extensive test cases by analysing requirements and code patterns. It can predict defects using machine learning. Detect visual inconsistencies across devices, and produce realistic, privacy-compliant synthetic test data. Additionally, Agentic AI can autonomously maintain and self-heal test scripts, adjusting their logic as underlying code or user interfaces change.
Furthermore, AI systems themselves need rigorous evaluation. Techniques such as red teaming, rainbow teaming, benchmarking, bias and ethics checks, and drift monitoring are essential to help promote AI’s reliability, fairness, and alignment with business objectives.
Human oversight is critical. While AI can scale testing and automate numerous tasks, critical thinking, risk assessment, and judgment cannot be fully delegated. Humans must guide, validate, and refine AI outputs to maintain both quality and trust.
Emerging Roles and Responsibilities
AI is reshaping professional roles. Developers are increasingly using AI by instructing machines through natural language rather than traditional programming methods. This shift has led to the emergence of new roles such as AI agent orchestrators, prompt engineers, QA specialists for autonomous systems, and governance leads who ensure ethical and auditable AI practices.
These roles are essential for maintaining human oversight. Developers and testers must experiment, validate, and continuously refine AI outputs while being cautious not to rely too heavily on AI.
Trust in the Age of the Apex Predator
As with any apex predator, AI has changed the rules of the game. Software once “ate the world” by making systems programmable. Today, AI “eats software” by making it autonomous, capable of creating, modifying, and deploying autonomously. In this new environment, speed is no longer the ultimate measure of success; trust is. Systems may move fast, but without rigorous QA, ethical oversight, and human judgment, they may not be reliable, accurate or ethical.
The new apex predator demands adaptation. Organisations navigating this AI-driven era must embrace automation and innovation, but pair it with strong quality practices, governance, and continual human oversight. Only by combining these elements can companies ensure their AI systems are not only fast and efficient but also dependable and aligned with business objectives. In a world of autonomous systems, trust is the ultimate competitive advantage.
Tom Lanaway is Head of Innovation at Connective3, a global brand & performance marketing agency. He leads a team building AI-powered marketing measurement and marketing intelligence tools.
SHARE THIS STORY
Most businesses are asking the wrong question about AI. They’re asking, ‘Which AI tool should we use?’ They should be asking: ‘Can our people actually think with AI?’
I run an innovation team at a marketing agency. We’ve spent the last two years building AI into everything we do, including measurement, content, strategy, and automation. We’ve got lots of tools, 18 different products to be precise.
Below is what I’ve learned. But the tools aren’t always the bottleneck; sometimes the skills are.
The Tennis Racket Problem
A colleague put it perfectly recently: “AI is a tool. Think of it as if you’ve got a smart assistant sat there. But it’s saying, I’m going to give you the best tennis racket, now go and play in a Grand Slam.”
That metaphor stuck with me because it captures something the artificial intelligence hype cycle keeps missing. We’ve convinced ourselves it democratises everything. That anyone can now do anything. That the barrier to entry has collapsed. And there’s truth in that, but it’s incomplete. The barrier to access has collapsed, but the barrier to effectiveness hasn’t. Give someone GPT-4, and they can generate text. Give them the best tennis racket, and they can hit a ball. But the gap between hitting a ball and playing at Wimbledon is still vast. Most organisations are stuck in that gap, wondering why their AI investments aren’t transforming anything.
Three Skills That Aren’t Always Present
When I look at where teams struggle and where I see the same patterns across other businesses, three specific competencies keep showing up as gaps:
1. Problem Decomposition
Not everyone knows how to break down complex work into chunks that AI can help with. This sounds simple, but it isn’t. Most people approach AI with whole tasks such as ‘Write me a marketing strategy’, ‘Analyse this data’ Or ‘Create a campaign’. AI will then produce something, but it’s usually mediocre, because the person hasn’t done the harder work of understanding which specific parts of that task AI is good at, and which parts need human judgment. The skill isn’t using AI; it’s knowing what to give it. Someone who is brilliant at their job but can’t decompose problems will get worse results from AI than someone more junior who understands how to break work into the right pieces.
2. Output Assessment
How do you know if what AI gives you is good? This is where intuition becomes essential and it’s also where the ‘AI replaces expertise’ narrative falls apart. You need domain knowledge to evaluate AI output. You need enough experience to feel when something’s off, even if you can’t immediately articulate why. You need the pattern recognition that comes from years of doing the actual work. Artificial Intelligence doesn’t replace that intuition; it requires it. The best AI users I’ve observed aren’t the most technical; they’re the ones who’ve built up enough expertise in their field to quickly assess whether AI output is useful, directionally correct, or completely off base. They know what good looks like, so they can recognise it when they see it, or notice when it’s missing.
3. Articulation
Can you clearly express what you really want? This is the unglamorous core of the whole thing. Some people struggle to articulate their requirements to other humans, let alone to AI. We’ve all sat in meetings where someone spends 20 minutes explaining what they need, and you’re still not sure what they want. AI makes that problem worse. The skill isn’t ‘prompt engineering’ in the technical sense; it’s the much older skill of clear thinking and clear communication. If you can’t articulate what you want specifically, precisely, with the right context and constraints, you won’t get useful output from AI or from anyone else.
The Uncomfortable Implication
Here’s what this means for how businesses should think about AI investment:
Stop leading with tools: Most organisations have tool fatigue already. Another platform, another integration, another training session on which buttons to click. It’s not working.
Start with the human work: Before asking ‘What AI should we use?’, ask ‘Can our people break down problems, assess output, and articulate requirements?’ If they can’t do those things well without AI, they won’t do them well with AI either.
Invest in the skills, not just the access: This doesn’t mean AI prompt engineering courses; it means developing clearer thinking, better problem decomposition, and sharper articulation. These are old skills, applied to new tools.
Accept that expertise still matters: The people who’ll use AI best are the ones who already know their domain deeply. AI amplifies competence; it doesn’t create it.
Connected Intelligence Isn’t About Connected Systems
I’ve spent a lot of time thinking about how different marketing channels and data sources connect and how you build intelligence across systems rather than in silos.
But I’ve come to think the more important connection isn’t between systems, it’s between human judgment and AI capability. The integration layer that matters most is the one between the person and the tool.
Get that wrong, and it doesn’t matter how sophisticated your AI stack is. Get it right, and even basic tools become powerful.
Hampshire Trust Bank (HTB) is using artificial intelligence (AI) to act faster on customer concerns. It is empowering its teams…
SHARE THIS STORY
Hampshire Trust Bank (HTB) is using artificial intelligence (AI) to act faster on customer concerns. It is empowering its teams to identify and respond quickly, whilst also meeting regulatory timeframes for handling complaints and supporting vulnerable customers.
Netcall: AI-Powered Sentiment
The specialist bank has worked with Netcall to deploy AI-powered sentiment analysis using Netcall’s Liberty Create platform. The solution reduces manual effort and improves operational efficiency by bringing customer emails from multiple mailboxes into a single interface. Incoming messages are automatically analysed to identify dissatisfaction, highlighting cases that may require faster intervention. This allows urgent cases to be prioritised, helping HTB to resolve issues before they escalate and improve the customer experience.
“Our AI-powered sentiment analysis solution rapidly processes vast amounts of email data. Its efficiency allows our team to focus on resolving customer enquiries and issues rather than sorting priorities. The streamlined process ensures swifter responses and better customer outcomes, upholding our reputation for exceptional customer service.” Ed Eames, Head of Customer Savings Operations at Hampshire Trust Bank.
The application was built by the Hampshire Trust Bank development team using Liberty Create. It worked closely with Netcall to integrate AI sentiment analysis into existing processes. Customer-facing teams were involved throughout to ensure the solution aligned with established workflows and regulatory requirements.
Customer Service Control
A key benefit of the approach is the level of control it gives internal teams. Keywords, sentiment thresholds, and classifications can be adjusted directly. This allows rapid refinement as customer behaviour changes or new regulatory considerations emerge, without waiting for development cycles.
“Liberty Create has enabled my development team to work with remarkable agility. The ability to rapidly create and refine applications to meet ever-evolving business needs has significantly enhanced our efficiency. This allows us to deliver a wealth of new features to end users and customers with speed. With the integration of AI, we’ve been able to advance our processes while ensuring exceptional customer service. Our Sentiment Analysis application launch is a prime example of this.” Trina Burnett, Head of Engineering at Hampshire Trust Bank.
The sentiment analysis system also supports automated and ad-hoc reporting. This provides a single source of insight into customer interactions and actions taken. This helps reduce manual effort, supports audit and compliance activity, and enables teams to continuously improve customer service operations.
“As scrutiny around customer experience and accountability increases across UK financial services, the ability to listen, adapt and respond at pace is becoming a defining capability for banks seeking to maintain trust and service standards,” said Alex Ballingall, Key Account Manager at Netcall.
“HTB’s approach shows how banks can use AI-driven insight practically. Turning customer communications into faster action without adding operational complexity,” Ballingall concluded.
About Netcall
Netcall is a leading provider of low-code and customer engagement solutions. A UK company quoted on the AIM market of the London Stock Exchange. By enabling customer-facing and IT talent to collaborate, Netcall takes the pain out of big change projects. It helps businesses dramatically improve the customer experience, while lowering costs. Over 600 organisations in financial services, insurance, local government and healthcare use the Netcall Liberty platform to make life easier for the people they serve. Netcall aims to help organisations radically improve customer experience through collaborative CX.
Patrick Cooney, CFO at Version 1, on why, in an AI-driven operating environment, financial discipline is more important than ever
SHARE THIS STORY
Over the last decade, digital transformation has become part of the CFO’s remit. As organisations invested in automation, cloud and data platforms, finance leaders were well placed to oversee spend, drive efficiency and ensure technology investments delivered measurable returns. As artificial intelligence (AI) moves from experimentation into the core of how organisations operate, that model is beginning to evolve. Primarily because AI-scale transformation demands a different balance of expertise.
A recent move by Coca-Cola illustrates this shift. The decision to take digital strategy out of CFO John Murphy’s remit and appoint Sedef Salingan Sahin as the company’s first Chief Digital Officer is not a rejection of finance-led transformation. It reflects a practical reality. While strong financial discipline remains essential, the architectural complexity and technical depth required to embed AI across an enterprise now go beyond traditional finance capabilities alone.
In all industries, AI is no longer confined to innovation labs or isolated pilots. It is increasingly embedded in how organisations operate, make decisions and manage risk. At Version 1, our earliest focus on AI was external: helping partners use AI to transform their own businesses. We have quickly turned the lens inward. Over the past quarter, we have accelerated the use of AI across our own finance and operational functions, implementing a wide range of practical use cases that fundamentally change how work gets done.
Some of these are relatively simple but have had a significant impact. Using AI to summarise documents, generate meeting notes or surface insights from large volumes of information has become normal and is already saving time across the organisation. Others are more structural. In finance, we are applying AI to areas such as accounts payable, accounts receivable and general ledger reconciliations, where large datasets and repetitive processes create natural opportunities for automation and acceleration.
We are also rethinking reporting itself. Rather than manually producing variance analyses each month, we are developing standardised prompts that allow AI to highlight key trends, explain deviations from budget and surface insights that would traditionally take hours to compile. These are not abstract efficiencies. Rather, they directly affect the speed, quality and value of financial decision-making.
What is striking though is the pace of change. Even over the past few months, usage has increased exponentially as people find new ways to integrate AI into their daily work. This is no longer an optional experiment. AI is reshaping how organisations function from the inside out.
Modern CFOs Deliver Stewardship and Governance
One of the biggest challenges CFOs face with AI is that traditional ROI models struggle to capture its true impact. Unlike earlier waves of digital transformation, AI does not deliver value solely through cost reduction or headcount optimisation. Increasingly, its value lies in better planning, faster decision-making, improved risk management and higher-quality outputs.
I see this clearly in how we use AI for planning. Recently, we fed a combination of internal data, previous plans and external consultancy material into a large language model and spent time crafting a detailed prompt. The output was a first-pass design for a major simplification programme (including workstreams, resourcing requirements and sequencing) that would previously have taken weeks to develop.
It is worth noting that this new process didn’t replace human judgement – it dramatically accelerated it. We are using similar approaches to shape annual finance priorities, drawing on historic plans and organisational context to generate structured, actionable starting points. This kind of value is real, but it does not always show up neatly in short-term financial metrics.
At the same time, the risks associated with AI are increasing. Model drift, regulatory scrutiny, data security and vendor dependency all carry financial implications. This is why governance matters as much as innovation. At Version 1, we have put formal structures in place, including an AI oversight committee that reviews and approves new tools, ensures appropriate controls are in place and sets clear boundaries around responsible use. We tightly manage which platforms can be used and how data is protected, recognising that public, uncontrolled tools pose unacceptable risks in an enterprise environment.
This combination of accelerating value and growing risk is precisely why ownership models are changing. Many CFOs continue to play a leading role in digital transformation, with research showing that around three-quarters of finance leaders now prioritise digital strategies at the highest levels of the organisation.
People Remain at the Heart of AI Adoption
As AI scales, the CFO’s role is shifting from delivery ownership to strategic stewardship. Finance leaders are uniquely positioned to connect technology ambition with financial reality, ensuring AI investments are governed properly, aligned to business outcomes and measured over time.
This aligns closely with how we think about our own operating model at Version 1. We use what we call a ‘strength in balance’ business model, built around three equally important pillars: customers, people and a strong organisation. That final pillar includes financial performance, risk management, cybersecurity and governance, all areas that become more critical, not less, as AI adoption accelerates.
People are central to this conversation. AI inevitably raises questions about job impact and cost optimisation, and organisations have a responsibility to approach this responsibly. That means clear communication, strong change management and treating people fairly where roles evolve. It also means investing in training and enablement. We have rolled out organisation-wide AI training focused on responsible use, and we are developing a network of AI champions with deeper skills who can identify and build use cases without relying solely on central teams.
The most effective model I see emerging is a shared one. Specialist digital leaders focus on building and embedding AI capabilities at scale. CFOs retain accountability for financial discipline, data governance and value realisation. When these roles work in partnership, organisations are far more likely to capture the value they expect from AI.
Financially Guided Value Delivery
As AI becomes a baseline capability rather than a differentiator, debates about who ‘owns’ digital strategy are becoming less relevant. The more important question is how organisations ensure AI investments deliver measurable, sustainable value. For CFOs, AI is now undeniably a balance sheet issue.
Investment in the latest technology affects cost structures, risk exposure, governance and long-term resilience. Those who engage proactively, shape governance and demand disciplined value creation will help their organisations unlock lasting advantage. Those who remain passive risk inheriting complexity, cost and compliance challenges that are far harder to unwind later.
In an AI-driven operating environment, financial discipline is not diminished. It is more important than ever.
Gregory Mostyn, CEO and co-founder of Wexler, on why the era of generalist AI tools is over, and how the future will focus on high-precision AI designed for specific industries
SHARE THIS STORY
For decades, the UK’s professional services sector, including areas such as Law, Insurance, and Wealth Management, has argued that its business value is locked in its access to proprietary data and the specialised labour required to navigate it. Investors, lured by the moat of institutional knowledge, priced these companies accordingly. However, the first quarter of 2026 has seen significant AI disruption within the professional services market. The catalyst wasn’t a single event, but rather a move by foundational model providers that turned the industry’s most defensible assets into commodities.
When Anthropic launched its specialised legal AI plugin, OpenAI integrated a real-time insurance underwriting engine directly into its interface, and Alturist Corp automated bespoke tax strategies, the market reacted harshly. As professional services titans such as RELX, MoneySuperMarket, and St James’s Place saw their share prices decline by more than 10% in a matter of hours, the message became clear: the era of treating AI as a ‘future risk’ is over.
The market has been awoken to the fact that foundational AI models are no longer just plugins or nice ‘add-on’ tools; they are competitors. The move by foundation-model providers into professional services – like the legal sector – is not a one-off shock, but rather an inevitability.
The Proliferation of Information
Historically, a law firm’s competitive advantage was its access to information – repositories of case law, proprietary research, and historical contracts. Investors and clients valued these companies on the assumption that this data constituted an impenetrable barrier to competitors. Before AI entered the mainstream, the cost of extracting actionable information from thousands of pages of data required a small army of junior associates and hundreds of billable hours.
In 2026, that moat has mostly evaporated. Recent benchmarks show that frontier models now achieve 80% accuracy on complex documents, compared with the 71% average of a human associate. More importantly, they do it at a fraction of the cost. It is now estimated that the inference cost for a system at the level of GPT-3.5 dropped by more than 280-fold between November 2022 and October 2024. It’s predicted that UK law firms will reduce their chargeable hours by 16% through the implementation of AI.
The narrative that AI would be able to handle only ‘low-level’ tasks, such as NDAs or simple contract summaries, has all but evaporated. Anthropic’s move into high-stakes litigation support validates this trend.
AI – From Swiss Army Knives to Scalpels
An error made by many law firms when AI became entrenched within the market was to treat it as a ‘plug-in’, a nice-to-have built onto existing internal software. Many adopted general-purpose tools, often referred to as ‘Swiss Army knife’ solutions, that covered the breadth of legal work but lacked the precision, jurisdictional nuance, and risk-weighted requirements for high-stakes professional services.
The 2026 market reaction highlighted the needs of a ‘scalpel’ approach – those that go deep in a specialised vertical within a legal workflow. For example, instead of a junior associate spending billable hours searching through case files to establish the facts of a case, they could use a ‘fact intelligence’ platform that can automate that process into minutes, whilst increasing accuracy by 95% versus 78% for human reviewers and up to 90% savings in large-scale litigation. The market is no longer rewarding firms for having information. Rather, it rewards those who can apply it at the lowest possible cost and friction.
Reallocating Capital Across Professional Services
We’re already seeing investors withdrawing from the traditional software market and reallocating that capital into specialised AI firms. However, the risk for legacy players is that they are being disrupted from both ends. From the bottom, they are losing the efficiency game to generalist foundation models from companies such as OpenAI and Google, which are commoditising the ‘knowledge’ aspect of professional services, including basic advice and contract drafting. At the top, they are losing the expertise game to specialised firms that use AI as a precision instrument; their overhead would be lower than that of a traditional Magic Circle firm, allowing them to undercut prices while maintaining profit margins.
The result is a massive reallocation of capital. Investments into vertical AI (AI built for one specific industry) are expected to surge to $115 billion by 2034. The market no longer bets on labour with tools, but on autonomous workflows. Investors have realised that the value lies in the middle layer – the software that sits between a general foundation model and a specific industry’s needs.
Innovation or Obsolescence
So far, the first market fluctuation of 2026 has taught us that you cannot outrun new technologies. To survive, firms must stop treating AI as an add-on and treat it as a foundation for their core business infrastructure.
For UK professional services, the choice is no longer whether to adopt AI, but whether they can evolve quickly enough to avoid becoming the training data for companies building foundational models. The firms that remain in 2030 will recognise that the competitive landscape has changed. You’re not just competing with your peers, but with the compute cycles of the world’s most powerful AI labs.
The era of generalist AI tools is over, and the future will focus on high-precision AI designed for specific industries.
Jack Bingham, Regional Director of Digital Native UK, Ireland & South Africa, Confluent on how data, treated properly, compounds in value to drive digital disruption
SHARE THIS STORY
When I talk to founders and tech leaders, one question seems to consistently come up: what separates today’s disruptors from the last decade’s? In 2010, being cloud-first was what made investors sit up and take note. In 2026, it will be streaming-first.
I’ve spent the last year or so working closely with companies that are, quite literally, building their businesses in real time. For them, real-time capability isn’t a department or a layer that supports the business. It is the business. The acid test is simple: how quickly can you capture a critical event – a payment, a login, a failed delivery – and respond with the next best action? That focus shapes how they build products, structure teams, and think about innovation.
Here’s what I’ve learned from them:
Lesson 1: Data is a Product, Not a By-Product
Many traditional companies still treat data as something to collect, store, and analyse later. The new generation of businesses, on the other hand, treats it as a reusable, governed product that everyone can access. When it’s built and shared this way, teams stop rebuilding the same foundations for every new use case. They move faster because they’re working from a single, trusted view of the truth, shortening product cycles, speeding up iteration, and spending more time solving problems that matter.
That mindset, rather than the size of the tech stack or the number of engineers, is what sets disruptive businesses apart. In these organisations, technology, data, and business strategy move in lockstep. Decisions aren’t passed up and down hierarchies, they’re made by teams who understand both the data and the customer problem in front of them.
When you can trust your data and respond in real time, innovation stops being a department. It becomes a reflex.
Lesson 2: Real-Time isn’t a Feature, it’s a Foundation
A few years ago, one of the world’s largest supermarket chains realised it didn’t have a single real-time view of its inventory. Without that visibility, omnichannel experiences were impossible. Once it shifted to a streaming architecture, every transaction became a live event that updated stock, triggered supply chains, and even made it possible to get your groceries delivered straight to your kitchen fridge – coordinated through live inventory data, smart home devices, and real-time security feeds.
That’s the practical power of streaming: it connects what happens in your business to what should happen next so you can provide products and services that take customer satisfaction to a whole other level. Real-time data stops being a reporting tool and becomes the foundation of every decision, interaction, and innovation.
I often ask businesses what they would do differently, if they knew the state of every event in their organisation. The most forward-thinking companies already have the answer. They’re using streaming to turn business events into reusable building blocks, creating new experiences by connecting the data they already have in smarter ways.
Lesson 3: Culture is the Multiplier
Being streaming-first is only half about architecture. The other half is attitude. The best digital enterprises don’t wait for permission to experiment. They map their most important business events, align teams around them, and empower people at every level to react fast and learn faster.
And the difference is visible. Feedback loops are shorter. Structures are flatter. Failure is treated as information. This culture of continuous experimentation is why these companies can move at the pace they do.
We often run ‘Event Storming’ workshops with teams to map their critical business events. The idea is to create alignment – getting people from engineering, product, and operations to agree on what really matters and how those moments connect. That process reveals a lot.
Digital disruptors go beyond simply deploying streaming architectures. They build streaming mindsets. Leadership plays a crucial role here: data must be treated as a strategic asset. If it isn’t up top, it won’t be anywhere else in the organisation either.
Lesson 4: Streaming and AI will Converge
AI is only as good as the data you feed it. Unfortunately, most enterprises are still feeding it yesterday’s data. Streaming-first companies already know this. They’re building intelligent data pipelines that give AI the context it needs to make decisions in real time.
That’s how the next generation of innovators will pull ahead: not by having bigger models, but by having cleaner, faster, more connected data. Streaming is what will let AI move from reactive to predictive… and from predictive to autonomous.
Too many organisations are cutting investment in data while pouring money into AI projects. But AI without quality data is just expensive guesswork. The companies doing this well understand that data has to be a product in its own right. And when business and technology teams design around that shared understanding, innovation follows naturally.
Lesson 5: The Mindset of the Next Disruptors
If I were starting a company tomorrow, I’d look closely at the critical events that run my business. I’d then make sure I had a way to capture those in the stream, make them reusable, and build every product and process around them.
When your business can see and act on what’s happening in the moment, you gain something no traditional architecture can give you: time. And in the next wave of disruption, that’s the only advantage that really matters.
If we look to who we can learn from in the coming months, it’s financial services and healthcare that are moving the fastest. Real-time fraud detection, patient monitoring, and risk management are becoming operational necessities – and these industries will set the benchmark for real-time data excellence.
Looking Ahead to 2026
By 2026, I don’t think we’ll talk about ‘real-time’ as a differentiator. It will simply be how modern businesses operate. Batch systems won’t disappear, but they’ll coexist within a single, streaming-first platform that delivers data whenever it’s needed.
Once every process can react instantly, the question then becomes: can it anticipate? Can it learn? That’s where AI and streaming meet and where we move from reactive to autonomous enterprises that not only respond to the present but adapt to what’s coming next.
Data, treated properly, compounds in value. The decisions you make with it become faster, sharper, and more confident. The companies that understand this will be the ones still leading when today’s titans look like yesterday’s news.
Dan Nichols, Chief Technology Officer at virtualDCS, on why cloud resilience in the financial services sector hinges on shared accountability and an assume-breach philosophy
SHARE THIS STORY
A powerful catalyst for transformation, the cloud is reshaping how organisations compete in the financial services sector. Beyond significant cost savings and flexibility, leaders are eager to unlock the potential of AI-driven insights, intelligent automation, and real-time business modelling. And, in a space governed so strictly by data sovereignty and privacy policies, the cloud’s ability to localise, encrypt, and control data has made it a key enabler of compliance and customer confidence.
But as threats become more frequent and sophisticated – with attackers now targeting shared platforms and partner supply chains – organisations can no longer rely on their own defences alone. For true digital resilience, shared accountability, collective readiness, and clear governance across every cloud touchpoint are equally non-negotiable.
All Eyes on the Money
The industry sits at a valuable intersection of data, technology, and finance. A combination that makes it uniquely attractive to attackers. It holds some of the world’s most sensitive data, directly underpins the flow of global capital, and operates through deeply complex and interconnected systems. With every integration increasing the risk of exposure. Ultimately, the attack motivation is as simple and relentless as it is in most sectors: monetary gain. Cybercriminals target institutions precisely because of the value at stake and the speed at which disruption translates to loss.
How the Threat Landscape is Evolving
Ransomware groups may see insurers and payment providers as high-yield targets. They understand even seconds of downtime can induce multi-million pound losses. Under pressure to protect customer trust and avoid regulatory penalties, some firms may choose to pay in order to restore their service quickly. This dangerous perception only encourages repeat targeting and paves the way for damage to spread even further. Yet it remains a common response tactic among many.
At the same time, the rise of supply chain and third-party attacks has made it possible for criminals to bypass even the most well-defended cloud environments. By exploiting shared platforms, managed service providers, and cloud-hosted applications, perpetrators can move laterally across multiple organisations at once, amplifying both the reach and impact of their attacks. In other words, infiltrating one vendor’s weakness can cripple an entire network in one carefully coordinated strike. And, since some firms may overlook the cloud’s shared responsibility model – presuming end-to-end security sits solely with their cloud provider – multiple blind spots can inevitably emerge, creating easy openings to exploit.
In an environment where boundaries blur and dependencies multiply, traditional perimeter-based defences are no longer enough. Hybrid and multi-cloud infrastructures demand continuous visibility, faster detection, and coordinated response across every partner and provider. The goal is not simply to prevent breaches, but to withstand and recover from them collectively. It’s about recognising that in today’s ecosystem, no financial institution is secure in isolation.
Inside the Ransomware Economy
Evolving beyond the scattergun attacks of the past, ransomware now operates as a professionalised, profit-driven ecosystem, where malicious actors collaborate, trade intelligence, and lease attack tools much like legitimate software vendors. The rise of ransomware-as-a-service (RaaS) has even lowered the barrier to entry, giving less skilled affiliates access to ready-made payloads and automated encryption kits in exchange for a percentage of the ransom.
What makes it especially destructive is the precision and psychology behind the attacks. Rather than randomly striking, attackers conduct weeks of reconnaissance – learning behaviours, studying employee hierarchies, and identifying systems most critical to operations. They often infiltrate through phishing emails or compromised credentials, quietly moving laterally through the network to gain elevated access. Once embedded, they disable defences, exfiltrate sensitive data, and target backup repositories before finally encrypting production systems.
At that point, the goal shifts from technical control to financial coercion. Victims are locked out of their systems and presented with a ransom note demanding payment, sometimes in cryptocurrency, in exchange for a decryption key. Increasingly, the threat includes public exposure of stolen data – a tactic designed to pressure leadership into paying to protect their reputation and customer trust. Even when ransoms are paid, recovery is rarely clean: data may be incomplete, corrupted, or resold on the dark web, and repeat targeting is common once an organisation is identified as a payer.
It’s this blend of stealth, strategy, and human manipulation that makes ransomware so difficult to defend against. By the time the encryption begins, attackers have already spent weeks ensuring recovery options are limited. This background isn’t designed to scaremonger, but to highlight why resilience must start long before an attack ever reaches the endpoint.
The Foundations of Ransomware Resilience
Ransomware resilience isn’t achieved through a single product or policy – it’s the outcome of strategic, technical, and cultural alignment. Financial institutions, in particular, must approach it as a continuous process of readiness: Anticipating compromise, containing impact, and restoring normality quickly and transparently:
Assume-Breach Philosophy
The first step is shifting from a defensive mindset to an assume-breach philosophy. In practice, this means recognising that even the most sophisticated systems can and will be breached – and building architectures and response strategies designed to limit damage when this happens. It’s a pragmatic approach, grounded in the reality that attackers are increasingly sector agnostic. No organisation is too small or too secure to be targeted, but the financial sector remains a favourite because it offers both high disruption value and potentially significant monetary reward.
Building meaningful resilience, therefore, demands layered defence and disciplined execution. The goal is to slow attackers down at every stage – detecting them early, limiting lateral movement, and ensuring business continuity when systems are disrupted. Behavioural analytics and continuous monitoring can surface and neutralise subtle anomalies that would otherwise go unnoticed – such as phishing, spear phishing, and malware, with email still the number one entry point for ransomware.
Zero Trust & MFA
Meanwhile, zero trust policies and multi-factor authentication methods add a second layer of protection, blocking unauthorised access even if credentials are compromised.
When incidents do occur, a well-practised response framework ensures action is fast and coordinated, minimising disruption across critical systems, with the ability to switch to secure replica environments to keep operations running while remediation takes place. Secure, immutable, air-gapped backups underpin it all, providing a safety net that guarantees recovery can begin from a clean and uncompromised state.
Human readiness is equally critical. Technology can contain an attack, but only people can recover from one effectively. Regular simulation exercises, incident rehearsals, and cybersecurity awareness training help teams respond calmly and cohesively, transforming response from reactive to instinctive. This operational maturity is reinforced by strong governance. Frameworks such as DORA, NIST, and ISO 27001 provide the structure to align technical teams, compliance leads, and executive decision-makers around shared resilience goals. When combined with skilled practitioners and clear accountability, they embed security into ‘business as usual’ – moving resilience from a strategy to a sustained organisational capability.
Why Multi-Layered Backup is Critical
When ransomware strikes, the speed and integrity of data recovery determine whether disruption lasts minutes or days – and whether the impact cascades through wider global markets. As the last and most decisive line of defence when every other control fails, it’s also fundamental to customer trust and compliance. Yet too often, backup is treated as a static safeguard rather than a dynamic resilience layer.
Since modern ransomware often seeks out and encrypts traditional backups first, a single backup copy or centralised repository is no longer sufficient. True resilience today depends on a multi-layered approach – combining offsite or cloud-diverse storage, immutable data copies that cannot be altered or deleted, and isolated environments to protect against lateral movement.
How frequently these backups are tested is equally important. Too often, financial institutions only discover weaknesses when recovery is already underway, at which point strategies can’t be magically strengthened, and it becomes a race against the clock to minimise downtime and reputational fallout. Regular, automated recovery testing changes that dynamic. It not only confirms that files can be restored, but provides verifiable assurance that systems come back online in the correct order, data dependencies remain intact, and teams have the muscle memory to act quickly and confidently when the worst happens.
The Power of Shared Accountability
In a digital economy so deeply interconnected, no organisation operates in isolation. This is especially true in financial services, where supply chains and service providers form the backbone of day-to-day operations. While this interdependence is a strength in many ways, it also means resilience is no longer defined by how well a single institution can defend itself, but by how effectively every partner in its ecosystem upholds their part of the security chain.
This is where shared accountability becomes critical. It recognises that cloud providers, managed service partners, and financial institutions each have distinct but complementary roles to play in securing data, systems, and infrastructure. When accountability is clearly defined – and when partners collaborate rather than operate in silos – visibility improves, incident response accelerates, and the risk of systemic failure decreases.
Shared accountability also extends beyond contractual obligation. It’s about building a culture of collective readiness: sharing intelligence, rehearsing joint incident scenarios, and supporting smaller or less-resourced partners to raise their security baseline. The result is a unified entity capable of anticipating, absorbing, and recovering from disruption together.
Looking Ahead
To view cyberattacks as inevitable might seem pessimistic to some, but it’s an unfortunate truth that no amount of investment can eliminate risk entirely. In an era where threats are growing in both scale and sophistication, readiness becomes the true differentiator – particularly in such a high-stakes sector. For financial institutions, that means embedding security into culture, strengthening connections across supply chains, and continually testing their ability to withstand and recover as a united ecosystem. Only then can resilience become a strategic advantage rather than a defensive necessity, and unlock the cloud’s transformative potential with absolute confidence.
Michael Ault, Country Manager at integrated payments specialists myPOS, offers strategic advice for SMEs looking to scale through digital transformation and diversification
SHARE THIS STORY
Scaling a small business is one of the most rewarding, yet complex journeys for any entrepreneur. While growth brings opportunities for greater reach, higher revenue, and stronger market presence, it also demands foresight, discipline, and the ability to manage risk strategically. Securely integrating new technology is the main obstacle for 47% of SME’s, even though 76% of these businesses intend to expand their IT investment. This underscores a key point of tension, as many businesses want to grow through digital transformation but struggle to do so securely and sustainably.
The business landscape continues to evolve with changing customer expectations, technology, and economic conditions. For UK SMEs, the key to long-term success lies in achieving growth but also in building resilience. Sustainable scaling comes down to three principles: embracing technology pragmatically, diversifying intelligently, and investing in people and partnerships that strengthen resilience.
Leveraging Digital Transformation
Digital transformation is the foundation of business growth, especially for small business. Cloud-based solutions, automation, and data analytics help to streamline operations, reduce inefficiencies, and create better customer experiences. However, transformation must be purposeful, not performative.
The smartest approach is to scale technology investment incrementally, integrating flexible, modular systems that evolve with business needs. This approach not only lowers risk but also helps ensure digital maturity evolve over time. When SMEs use modular, cloud-based technology, operations run more smoothly and changes can be effectively analysed. Ultimately, resilience is not built through one-time upgrades but through a culture of continuous digital evolution.
Diversifying Revenue Streams
Depending on a single product, service, or market leaves a business vulnerable to sudden changes in demand. Diversification, when guided by customer insight and data can turn volatility into opportunity. Expanding into online sales, introducing subscription models, or targeting fresh customer segments can make income streams much more stable and sustainable.
At myPOS, we know that even simple changes based on data, such as adding additional payment options or tapping into cross-border e-commerce, can help cash flow and protect against market shocks. The goal of technology is to mitigate specific challenges without adding layers of complexity.
Investing in Employee Development
Your people are pivotal to your ability to grow as a business; empowered teams are the engine of sustainable scale. A team that feels supported and motivated will bring fresh ideas, adapt to challenges, and push the business forward. Investing in training, mentoring, and development opportunities builds skills that pay back in the form of innovation and improved performance.
In fast-changing industries, having employees who are confident in learning and adapting can make the difference between struggling through disruption and taking advantage of it. Equally, strong partnerships extend this resilience beyond the organisation. Building resilience at the team level creates resilience for the whole business, so fostering a culture of continuous learning and celebrating employee contributions is key to maintaining motivation.
Focusing on Financial Health and Flexibility
Financial resilience underpins sustainable growth. Scaling often requires upfront investment, and without healthy cash flow or reserves, opportunities can be lost. Monitoring income and expenses closely, cutting unnecessary costs, and preparing for seasonal fluctuations gives businesses more control.
Having flexible financing options, like credit lines, small business loans, or even crowdfunding, provides a level of agility. Instead of being caught off guard by unexpected challenges, businesses with financial flexibility are positioned to respond quickly and strategically.
Financial management software can make it easier to track performance, spot issues early, and forecast future needs. When you can see your finances in real time, you can make proactive, data-driven decisions instead of waiting for problems to happen. In markets that change quickly, this kind of financial management helps small firms plan with confidence, stay flexible, and establish a stronger base for long-term growth.
Prioritising Customer Relationships and Feedback
Your customers are not just buyers; they are advocates, sources of insight, and the foundation of repeat business and brand loyalty. Businesses that scale successfully often place customer relationships at the heart of their strategy by actively gathering feedback, responding quickly to issues, and personalising interactions, which shows customers they are valued.
This loyalty becomes a form of resilience. In periods of uncertainty, a base of satisfied, returning customers provides more stability than constantly chasing new ones. Successful businesses use CRM tools to track customer preferences and automate follow-ups so no opportunity to strengthen a relationship is missed.
Building Strategic Partnerships
Partnerships can accelerate growth while also spreading risk. Working with other businesses, organisations, or influencers can provide access to new audiences, shared expertise, or additional resources. Collaboration can also create opportunities for joint marketing, co-branded initiatives, or innovative product and service offerings.
In times of uncertainty, strong partnerships act as a support network. By aligning with others who share your values and vision, you create opportunities that are mutually beneficial and more resilient than going it alone. It is important to find partners whose goals and audiences complement your own for the best long-term impact.
The next stage of small business success will be defined by resilience rather than speed, the ability to adapt, recover, and continue to create value in the fact of uncertainty. For SMEs, this means developing adaptable growth plans that include flexible technology, diverse models and empowered employees.
Jan Van Hoecke, VP AI Services at iManage and a highly experienced computer scientist with a passion for technology and problem-solving. on navigating the AI landscape for success in 2026
SHARE THIS STORY
The AI landscape faces a number of big shifts in 2026. Agentic AI will undergo a reality check as enterprises discover the gap between marketing hype and actual capabilities, while organisations will go through a mindset change from treating AI hallucinations as crises to managing them, acknowledging the inherent limitations of the technology. There will also be a shift in how data will be structured in AI systems, to help the move from just finding facts (“what”) to understanding reasons (“why”). Middleware application providers will face new challenges, as those vendors controlling both platforms and data will become more influential. Finally, standardised AI chat interfaces will evolve into smarter, dynamically generated, task-specific user experiences that adapt to immediate needs.
Agentic AI Reality Check
2026 is the year when agentic AI will get a reality check, as the gap between marketing promises made in 2025 and their actual competencies will become starkly visible. As enterprise adopters share the mixed successes of agentic AI, the market will begin to differentiate between true autonomous agents and the clever workflow wrappers.
Currently, many products promoted as AI agents are, in reality, rigidly programmed systems that simply follow predefined paths. They cannot independently plan or adapt in real-time to accomplish tasks. The current evolution of AI agents closely resembles the development of autonomous vehicles: early self-driving cars could only maintain lane position by relying strictly on preset instructions, and likewise, today’s AI agents are limited to executing narrowly defined tasks within established workflows. True autonomy, where AI agents can dynamically perform and solve complex problems better than humans and without human intervention, remains, for now, an aspirational goal.
AI Hallucination Goes from Crisis to Management
In 2026, the AI hallucination crisis will reach a critical juncture as organisations realise they must learn to coexist with the current fundamentally imperfect technology – until a new technology comes into play that can effectively address the issue. The focus will shift from AI hallucination ‘crisis’ to management.
As the industry deliberates who carries the liability for AI’s mistakes and inaccuracies – the tool makers or the users – enterprises will stop waiting for vendors to solve the problem and take matters into their own hands. They will adopt a variety of pragmatic risk mitigation strategies – from double and triple-checking work, and enforcing human oversight for high-stakes decisions, to taking hallucination insurance policies.
Major model builders acknowledge that current foundational LLM technology cannot eliminate hallucinations and ambiguity through incremental improvements alone. New technology is needed. Until then, and perhaps with the realisation that a technological breakthrough is years away, users will start driving the hallucination conversation – both by building systematic defenses within how they use AI, and forcing vendors to accept shared responsibility through better documentation and clearer model limitations.
The Next Evolution in AI Data Architecture Lies in a Shift from “What” to “Why”
There will be a fundamental shift in how data is structured for AI systems, driven by the limitations of current approaches in answering complex questions. While Retrieval Augmented Generation (RAG) has proven effective at locating information and answering “what” questions, it struggles with the deeper “why” and “how” inquiries.
This limitation stems from RAG’s flat-file architecture, which excels at locating information but fails to capture the complex interconnections and relationships that underpin meaningful understanding and knowledge, especially in specialised domains like legal and professional services information.
The solution lies in AI-driven autonomous structuring of data. These systems will be better placed (than humans) to reveal critical relationships across multiple data points at scale, also highlighting the contextual dependencies essential for answering the “why” and “how” questions effectively.
Consequently, in 2026, with machines taking the lead, the method of structuring data will undergo a complete transformation, gradually eliminating the human role in creating structure, to reveal the business-critical interconnections across multiple data points.
Middleware AI Apps Squeeze
Given the essential link between data and AI, middleware companies that specialise in building custom applications layered on top of data platforms will begin to get pushed to the margins, forced to compete on niche features – while the core value of data and insight is captured by the platform owners. The true leaders will be those organisations that both own and manage their data, while also offering an AI-powered interface that enables users to interact with their data securely and efficiently, fully leveraging the capabilities of modern AI technology.
Shift to AI-generated, Task-Oriented User Interfaces
In 2026, the current traditional vendor-designed, standard AI chat-based user interfaces will transition to dynamically AI-generated task-specific user interfaces that adapt to users’ immediate needs. This represents a fundamental shift from standardised software – for example, where everyone uses identical Microsoft Word or SharePoint interfaces – to personalised, short-term user interfaces that exist only as long as the user requires them for a specific task.
This transformation will also address the critical pain point that users typically have – i.e, the crushing cognitive load of navigating bloated, feature-rich software. Instead of searching through endless menus in an overstuffed application like Excel, the user will simply state their goal – “Compare the Q3 and Q4 sales figures for our top 5 products and show me a chart” – and the AI will instantly generate a temporary, purpose-built interface – a “micro-app” – solely designed for that one single task.
In the context of dynamically generated user interfaces, both data storage and the creation of bespoke interfaces will be managed by AI. The AI organisations that will truly lead in providing such bespoke user interface-generating capability are those that possess and control their own data.
About iManage
iManage is dedicated to Making Knowledge Work™. Our cloud-native platform is at the centre of the knowledge economy, enabling every organisation to work more productively, collaboratively, and securely. Built on more than 20 years of industry experience, iManage helps leading organisations manage documents and emails more efficiently, protect vital information assets, and leverage knowledge to drive better business outcomes. As your strategic business partner, we employ our award-winning AI-enabled technology, an extensive partner ecosystem, and a customer-centric approach to provide support and guidance you can trust to make knowledge work for you. iManage is relied on by more than one million professionals at 4,000 organisations around the world.
