Banks warned AI design tools could accelerate weak digital products

Banks may soon be able to produce digital product concepts in minutes, but faster output will not by itself improve customer trust, adoption or long-term value, according to Alex Kreger, founder and chief executive of UXDA Financial UX Agency. In a Finextra community opinion post, Kreger argued that generative AI is reducing the cost and time needed to create financial interfaces while raising the risk that institutions build polished products around untested assumptions.

Kreger said AI tools are already helping product teams produce onboarding flows, wealth dashboards, lending journeys, conversational assistants and mobile interface concepts more quickly and cheaply. The commercial impact, in his view, is a shift in product economics: when prototypes and code can be produced rapidly, the scarce resource becomes judgment about which problems are worth solving.

The warning comes as AI capabilities are being added to design and product-development tools. Kreger cited Nielsen Norman Group research saying AI is being integrated into UI and UX tools, enabling teams to generate prototypes, wireframes, design systems and front-end or back-end code far faster than traditional workflows.

Execution speed versus customer understanding

Kreger said financial institutions have historically treated user experience as a late-stage presentation layer, applied after business rules, compliance needs, technology limits and stakeholder preferences have defined the product. That approach, he argued, can leave banks with attractive interfaces built over fragmented customer journeys or unresolved sources of friction.

AI does not resolve that weakness, Kreger wrote. If banks lack evidence about customer needs, behaviour and trust barriers, he said AI can help them generate more versions of the same flawed idea. In his assessment, the immediate availability of professional-looking prototypes may make internal approval easier while masking whether the product addresses a real customer problem.

Kreger pointed to U.S. Bank’s in-house Design AI Assistant, launched in March 2026, as an example of how large institutions are applying AI to design operations. According to his post, the tool helps experienced designers work faster while checking accessibility, brand, design and content standards, detecting issues, proposing compliant alternatives and applying fixes with one click.

He also cited Figma’s State of the Designer 2026 report, which found that 91% of designers said AI tools improve their designs, 89% said they work faster and 80% said collaboration has improved. Those figures support the case that AI is already changing design execution, although Kreger framed the greater challenge as governance rather than tool adoption.

Research risk and synthetic users

Kreger said the larger risk emerges when product teams use AI to replace customer research. AI can generate personas, pain points, journey maps and use cases, but he argued those outputs can become a closed loop if they are not tested against real customer observation.

He described AI-generated consumer panels, sometimes called synthetic users, as useful only if validated by human research. Without that validation, Kreger said banks may optimize products around plausible assumptions rather than observed behaviour.

Other research cited in the post points to pressure on design quality and differentiation. According to Designlabs’ The State of AI in UX & Product Design: 2026, more than half of respondents expressed concern about AI’s effect on design quality. Kreger also cited JD Power research saying the performance gap between the strongest and weakest banking apps and websites has narrowed to its lowest level, leaving customers with more consistent but less distinctive digital experiences.

Kreger argued that AI will move UX work earlier in the product process, toward hypothesis selection, customer psychology, trust formation, journey testing and strategic governance. In that model, smaller teams may coordinate networks of AI agents for design, coding, testing, documentation, deployment and monitoring, while humans define objectives and evaluate outcomes.

His conclusion was that banks should use AI to accelerate experimentation, accessibility checks and consistency, rather than as a substitute for research or product strategy. The advantage, he said, will accrue to institutions that combine AI speed with stronger judgment about customer needs, financial behaviour and digital business risk.

This story draws on original reporting from Finextra Research.