Financial Services AI Is Entering Its Harder Phase, Says Petr Dlouhý, Trask’s New FSI Director

What is the biggest technology shift you currently see in financial services?

PD: AI is clearly the strongest topic. But I would not describe it only as a new technology trend. AI is starting to reopen larger transformation topics that financial institutions have been dealing with for years: technical debt, back-office and IT transformation, and the way complex delivery is owned and managed.

Many banks and insurers operate legacy environments where every new change becomes harder, slower and more expensive. AI now makes it possible to address these long-postponed topics in a more structured way.

Many companies already use AI in everyday work. Is that enough?

PD: It is a useful start, but not enough. Today, many roles already use AI individually. Analysts, developers, testers or business users use it to speed up daily tasks. But the bigger opportunity is not individual productivity. It is the use of AI across whole processes and organizational units. That is where AI can start to change how a bank or insurance company actually works.

Where do you see the most relevant AI use cases for FSI clients?

PD: There are two important directions. One is efficiency. AI can help redesign back-office processes, call centre operations, mortgage processing, risk workflows or other areas where large teams handle complex, repetitive or document-heavy tasks.

The other is business growth. AI can support relationship managers, sales teams, branches and digital channels with better context, faster preparation and more relevant client interactions.

The question is not only: can AI make this process faster? The better question is: would we design this process differently if AI, data and integration were already available?

No integration, no AI scale

What prevents financial institutions from scaling AI faster?

PD: Very often, the bottleneck is not the AI model. It is everything around it. AI needs reliable data and access to the systems where real processes happen. Financial institutions with strong data foundations, APIs and integration architecture can move much faster because they can embed AI into real workflows. Others first need to solve data availability, quality and system connectivity.

So AI makes integration and data more important?

PD: Exactly. Data and integration have always been important enterprise capabilities. With AI, they become prerequisites. If AI is expected to support people, automate decisions or act inside workflows, it cannot sit next to the enterprise architecture. It has to be connected to it.

Can you give examples where this foundation already creates value?

PD: In financial services, integration and data are not abstract architecture topics. They directly affect business speed, customer experience and the ability to scale digital services.

One example is real-time customer engagement at MONETA Money Bank, where an event-driven platform enables the bank to react to customer behaviour within seconds and deliver more relevant offers. In one use case, this helped deliver a relevant offer within 7 seconds and increase marketing conversion by a factor of three.

Another example is the integration backbone for J&T Bank’s next-generation mobile platform, which connects more than 15 core banking systems, processes over 2 million messages a day and significantly reduces latency. In insurance, API Management for Vienna Insurance Group shows the same principle from another angle: secure, reusable and visible integration across digital and internal processes.

These are the kinds of foundations that make later AI adoption much easier, because the systems, data and integration patterns are already in place. They are the conditions that allow digital services, automation and AI to work at scale.

Legacy modernization is back on the table

You often mention technical debt. Why is it such a key topic now?

PD: Technical debt is one of the biggest barriers to change in financial services. Many systems have been evolving for years. They are critical, but difficult to change. Every new product, feature or regulatory requirement becomes more expensive and harder to deliver.

In the past, modernizing a large legacy system was often too risky or too expensive. These projects were postponed because the scope was simply too big.

How can AI change that?

PD: AI can help decompose the complexity. It can support analysis of legacy code, documentation, dependencies, historical incidents and system behavior. That helps teams understand what the system does, where the risks are and how modernization can be planned.

This is important because AI may not only accelerate modernization. In some cases, it is what makes modernization possible in the first place.

Does this also affect system outsourcing?

PD: Yes. Many clients struggle to maintain older technologies internally. The people who built the systems may no longer be available. Documentation may be incomplete. And keeping legacy knowledge inside the organization is expensive.

With AI-supported analysis and transition, we can take over complex systems faster and with more confidence. For clients, this means they do not have to solve every legacy technology internally. They can transfer responsibility to a partner that has the scale, delivery experience and tools to operate and modernize those systems.

FSI delivery starts with business understanding

Are expectations toward technology partners changing?

PD: Definitely. Technical expertise is still essential, but it is becoming the baseline. Clients increasingly expect us to understand the business context as well.

If we work on mortgages, payments, risk, digital channels or back-office processes, we need to understand how those processes work, not only the technology layer behind them.

The client is also not always only the IT department anymore. More often, we work with business units, tribes or specific business sponsors. They expect a partner who can explain how technology delivery supports business outcomes.

What does this mean for delivery?

PD: For complex transformation, pure capacity is not enough. Financial institutions need delivery models where both the client and the partner carry responsibility for the result.

AI makes this shift even more important. It can help teams deliver faster, analyze legacy systems more effectively and build reusable delivery capabilities across clients. But the value comes only when AI is connected to a clear delivery model, real business priorities and the process behind the technology.

From task automation to operating model

You mentioned back-office transformation. Why is this such a relevant area?

PD: Back office is a good example because it often combines complex processes, manual work, documents, decisions, exceptions and many system dependencies.

AI can help automate selected tasks, but the larger opportunity is to rethink how the whole process works. That may include AI agents, workflow changes, better data access, governance and integration with core systems.

In areas such as mortgage processing, operations or risk, the potential can be significant because even one process can involve large teams and high operational complexity.

Can you give a concrete example of this shift?

PD: A good example is MONETA’s GenAI Claims Automation for card payment claims handling. The solution uses GenAI to automate a large share of standard cases, provide near-instant customer responses and reduce manual work in the back office.

But the important point is not only the AI agent itself. To make this work in a bank, the solution has to understand the case context, connect to relevant systems, work with reliable data, respect process rules and keep human control where it is needed.

That is why even a focused AI use case quickly becomes an enterprise delivery topic. Integration, governance, security and operations all matter.

What role do AI agents and agent platforms play here?

PD: AI agents can become part of how work is executed. But in an enterprise environment, they cannot operate without control. Financial institutions need an agent platform that defines how agents access data, how they interact with systems, what they are allowed to do, how outputs are checked and how governance works.

This is where technology delivery and operating model design come together. It is not enough to build an agent. The institution needs a controlled environment where agents can be safely used across real processes.

7 FSI tech trends according to Petr Dlouhý

• AI is moving beyond individual productivity – from helping single roles to changing whole processes, teams and operating models.
• Data and integration are becoming AI prerequisites – reliable data, APIs and connected systems decide how fast AI can move into real workflows.
• Legacy modernization is back on the table – AI can help break down complex systems and make long-postponed modernization more manageable.
• Back office is becoming a major AI opportunity – especially in processes with documents, decisions, exceptions and high manual workload.
• Delivery ownership matters more than capacity – financial institutions need partners who understand the business process and take responsibility for outcomes.
• AI agents need enterprise control – governance, access rules, human oversight and security have to be part of the operating model.
• The banking interface may change next – banks need to prepare for AI agents acting on behalf of clients, with new demands on access, consent and trust.

The next banking interface may not be human

How could AI change the way end customers interact with banks?

PD: This is still emerging, but it is important. In the future, the primary interface of banking services may not always be a person using mobile or internet banking. It may also be an AI agent acting on behalf of the client.

That would change how banks think about access, APIs, data sharing, consent, authentication, authorization and security. The question will no longer be only how to improve the mobile banking interface, but also how to safely expose banking services to trusted agents, define what they are allowed to see or do, and maintain control, auditability and trust.

There is no final operating model yet, but financial institutions should already prepare technologically. Again, the foundation will be strong integration, data governance, consent management, identity and access management, and secure architecture.

How does this change the cybersecurity agenda for financial institutions?

PD: Security remains critical because financial institutions work with sensitive data and critical infrastructure. If AI agents become part of how clients interact with financial services, identity, access management, consent and auditability will become even more important.

AI can also make attacks more sophisticated, so the defensive side has to evolve. At the same time, topics such as post-quantum security and certificate protection are becoming more relevant.

For FSI institutions, AI adoption cannot be separated from governance, access control, explainability and security.

As you take over the FSI Director role, what kind of leadership does this next phase require?

PD: It requires technology depth, business understanding and delivery responsibility. In financial services, transformation is complex, regulated and highly operational — value only appears when teams can turn ideas into something that runs safely and at scale. So this kind of leadership has to stay close to delivery.

That is how I have grown at Trask over the past fifteen years: through delivery itself, leading teams, building long-term client relationships and taking responsibility for increasingly large parts of the business. It also shapes how I want to lead FSI now — give experienced people autonomy, lead through trust, and stay personally accountable for outcomes our clients can rely on.

About Petr Dlouhý

Petr Dlouhý is Industry Director for Financial Services & Insurance and Partner at Trask. In fifteen years at the company, he has built and scaled Trask's expert teams in enterprise integration and payments, growing them into some of its largest delivery practices in CEE. He has personally led many delivery engagements for banks and developed long-term strategic relationships with key banking clients.

As FSI Director, he now focuses on AI transformation and Trask's international growth in financial services.

‍