Endava: What Causes AI Adoption Gaps in the Finance Sector

Financial institutions express firm confidence in adopting agentic AI, yet they still avoid committing budgets.

As a result, Endava surveys one thousand senior leaders across nine countries and reports that 92% say their organisations stand ready to deploy autonomous AI systems, though only 36% set aside funding. 

Leaders do not question the technology or its uses. They point instead to missing infrastructure, governance processes and the capability required to run these systems at scale. 

This gap between ambition and implementation shapes the market they describe.

Matt Cloke, CTO at Endava, says: “Agentic AI represents a step-change in how financial services organisations operate and innovate. The opportunity is clear, but so is the responsibility.”

How agile methods create bottlenecks for autonomous systems

Three quarters of surveyed leaders say agile methods create bottlenecks. 

Agile describes a development structure built around short cycles known as sprints. 

Respondents say these sprints struggle to contain AI systems that operate on different timescales and adjust through autonomous learning. 

They describe a mismatch between fixed cycles and models that continually update.

Regulatory complexity and integration across ecosystems emerge as the areas where friction appears.

49% cite each of these as places where agile fails to keep pace. They say an autonomous AI system linking to multiple third party platforms and moving through cross border rules does not sit well inside a two week cycle. 

They point out that compliance layers such as anti money laundering checks and know your customer requirements vary across jurisdictions and demand processes that agile does not easily support.

Leaders still value agile. 86% say it remains useful for certain tasks. The challenge becomes working out when agile fits and when AI systems need different approaches. 

76% expect their organisations to require AI native operating models within two to three years. 

These models describe structures that form around autonomous systems rather than traditional development cycles. Nearly all surveyed, 94%, say this shift shapes competitive standing across the sector.

Yet only a small minority call themselves AI native today. 

16% across all surveyed markets use that term. Italy sits at twenty 5%, followed by the US at 24%. 

France stands at 6% and the UAE at 4%. The UK matches the global level at 16%.

Leaders describe this limited adoption as proof that most organisations still work inside legacy patterns while preparing for change.

Why agentic AI is expected to open new markets

More than 80% of leaders surveyed expect agentic AI to open new markets or create different revenue streams. 

Immediate uses focus on fraud detection, financial crime prevention and operational continuity. 

Respondents point out that banks lose billions of dollars every year to fraud and regulators continue to tighten anti money laundering and know your customer rules. 

Even brief system failures bring penalties and reduce customer trust. They say autonomous AI processes transactions at speed and spots anomalies faster than human teams.

They also speak about continuous operation. 

They anticipate twenty four hour service, faster product launches and more personalised customer interactions. 

Some say autonomous models will take on routine tasks without human involvement, with staff moving to work that requires human judgement. 

They frame part of this expectation as a defensive measure. Digital banks and fintech firms build their models around speed and customer experience, while traditional institutions carry older systems and deeper regulatory requirements.

How data privacy and regulatory uncertainty lead concerns

Data privacy, regulatory uncertainty and explainability appear most often when leaders discuss concerns about deploying agentic AI. 

Explainability describes the need to show how a system reaches a decision.

When an autonomous model blocks a transaction, flags an account or makes a lending judgment, customers and regulators want a clear explanation. Leaders say this becomes more complex when systems learn and adapt autonomously. 

Privacy also grows more difficult when AI models need access to information spread across jurisdictions with different rules.

Nearly half of surveyed organisations, 47%, say they build ethical guidelines into AI development. 

The same proportion say they implement transparency and explainability measures. Another forty 6% strengthen data privacy, while 44% set governance frameworks for AI use. 

Some 42% work to align agentic AI with regulatory requirements and thirty 7% train employees on responsible use. 

Respondents say governance failures risk turning the technology from asset to liability.

Matt says: “Our research shows that those who build AI-native operating models, backed by strong governance, will be the ones to lead the next era of financial services.

“At Endava, we’re already adopting this approach with Dava.Flow, our AI-enabled engagement lifecycle. 

“We know that success lies in adapting quickly, embracing multidisciplinary teams and balancing innovation with organisational health.”