The Evident AI Index 2025, a global benchmark that tracks how banks are adopting and scaling AI, shows that while most major institutions are investing in the technology, only a few are seeing meaningful results. The gap between leaders and laggards is growing, with U.S. banks ahead in talent and innovation. As companies focus on responsible AI and measurable returns, a small group of frontrunners is pulling ahead, leaving much of the industry playing catch-up.

The FR editors spoke with Evident co-founder and co-CEO Alexandra Mousavizadeh to discuss what the data reveals about the future of banking and AI.

If you had to sum up the 2025 Index in just one takeaway, what’s the big story here?

It’s going to become harder and harder to catch up. The data tells us that the leading banks are more than doubling the pace of AI progress and deployment. Very soon that translates into leaner cost structures, better client engagement — on both the retail and wealth sides — and lower cost of delivery.

Ten percent of AI adoption is the tool itself; 90% lies in complete business rewiring and data rewiring. That takes a long time. When you see the leading banks pull ahead, that’s not based on one year’s work; it’s five to seven years.

We’re at an inflection point. After two years of busy investing, centralizing, and selecting use cases, the leading banks are now moving into this next phase of rewiring. It definitely feels like we’re right in that tectonic shift right now.

What’s the most significant change you see compared to one year ago?

The use cases have changed a lot. Last year we saw proof-of-concepts or single-use-case tools announced with big fanfare. Now we’re seeing agentic AI use cases that rethink entire domains. We’re still a few years away from fully agentic workflows, but banks are starting to shift from individual use cases to full-domain transformation. You can see it clearly in the talent data — the ramp-up is significant among banks doing that.

The report shows that only a handful of banks are reporting measurable ROI from AI initiatives, even as adoption rises. What’s holding most institutions back from quantifying results? Is it a measurement problem, a data problem, or a maturity issue?

Measuring ROI is still in very early stages. Most banks are fragmented — progress is happening within lines of business, not centrally. The more mature banks have developed ROI frameworks, translating efficiency gains into dollar value. They’ve had internal reporting structures in place for a couple of years, at least.

Those banks are comfortable enough to say, “We’ve got an eye on most of it.” Their numbers are usually under-baked, which is fine — it’s about maturity and discipline. It’s also a signal to shareholders and potential recruits. Talent looks at whether a bank has a strategy, measures output, and is organized enough to make joining worthwhile. ROI measurement is ultimately a marker of maturity.

So it has a lot to do with how mature the use cases are — if they don’t have enough data from how they’re using AI, it’s harder to measure ROI.

Exactly. We’re still in the foothills. In a best-case scenario, I don’t think we’ll really see ROI hit the bottom line until 2027.

If the top ten banks are converting AI into significant value while others are still experimenting, what does that mean for competition? Are we seeing a concentration of digital advantage that could reshape the market?

If you’re too small or too late, you can’t compete. We’re already seeing major bifurcation. The leading banks are out ahead, and it’s going to be harder to catch up because the plumbing, reorganization, and governance rewiring take time. It’s taken the big banks a year or two just to steer the tanker in that direction.

You’ll see more concentration, though some smaller banks are exceptions — they’re digital-first and determined. It’s not purely a question of scale. It’s about prioritization and determination.

The report also notes that banks are ramping up on AI-related communications, including responsible-AI transparency. Do you see that as healthy disclosure, or is there a risk of overshooting expectations?

Banks are erring on the side of caution. You can’t release a use case without regulators taking interest, so banks are careful. You might see some announcements about things most peers already did a year ago, but that’s not exaggeration — it’s conservatism.

There’s hype, of course, because everyone’s trying to showcase performance, but I don’t think banks will overpromise. They’ll continue to disclose safely what they can stand by. Over time, we’ll see the focus move — from GenAI becoming business as usual, to agentic AI as the new frontier, and then toward the intersection of AI and quantum computing.

The report finds that U.S. banks dominate the AI-talent race. With tighter immigration and visa policies, how sustainable is that lead?

It’s a short-term, significant problem. Many U.S. bank employees hold H-1B visas. It won’t halt progress, but it’s a shame because it chips away at a major U.S. strength — the ability to attract global talent. Banks are already looking to hire in Asia or Europe to offset that.

As AI talent and data capabilities become table stakes, what’s going to separate the leaders from the followers in the next phase of banking AI maturity? Do culture and leadership agility matter more than scale?

Absolutely. The next phase is about execution. The first step was getting CEO and senior-tech leadership buy-in. Now the focus is on heads of lines of business — training them, embedding AI accountability into their KPIs, and pairing them with AI experts.

The leading banks are using a “two-in-a-box” model: pairing a line-of-business head with a PhD-level AI researcher or product manager. When that partnership works, progress is rapid. Embedding technical talent deeper in the organization is key to rewiring at scale.

What are some implications for community banks and credit unions?

Enjoying your read? Sign up here

The same dynamic applies. Credit unions and regional banks face competition too. It’s like when the internet emerged — those who said, “We’ll wait it out” disappeared. For smaller institutions, the key is cultural change. Get people using AI tools, find internal champions, and show what’s possible. It doesn’t require huge investment or turning into a tech company, but openness to innovation is essential. The big banks aren’t coming for them yet, but they still need to move.

Read the full report for more on Evident Insights’ benchmark rankings of banks’ AI efforts.