AI benchmarking and intelligence platform Evident published its latest AI Outcomes Report for Banking, analyzing 173 public AI use cases across 50 of the world’s biggest banks to assess where AI is delivering value – and where progress remains opaque.
While banks are rapidly scaling generative artificial intelligence (genAI) deployments, the report reveals that fewer than a third of these use cases currently disclose any outcome metrics. This signals a persistent maturity gap in how the industry measures return on AI investments.
The report shows that J.P. Morgan, DBS, RBC, BNP Paribas, and Société Générale have each disclosed business value from AI initiatives exceeding $500 million – whether realized or projected – reinforcing the case for clearer ROI attribution across the sector. More transparent outcome measurement is key to unlocking the next stage of enterprise AI maturity, from building investor confidence and regulatory trust, to benchmarking success internally and focusing investment on what’s working.
“Many banks are still in the early stages of calculating AI ROI and hesitate to disclose impact assessments due to competitive sensitivity or regulatory scrutiny,” said Alexandra Mousavizadeh, co-founder and CEO of Evident, in a statement. “However, reporting use case outcomes is a strategic signal that AI is no longer an experimental cost centre, but a measurable driver of growth, profitability, and competitive advantage. Greater transparency around AI outcomes and more open reporting is needed to create industry-wide momentum and drive to more successful AI adoption.”
Key findings from the Outcomes Report include:
- Banks announced twice as many AI use cases in 1H 2025 (vs. 2H 2024); more than half of these were focused on genAI applications
- Only 30% of use cases report outcomes: Despite the surge in activity, less than a third of use cases include metrics on efficiency, revenue, risk reduction, or customer impact
- Among those that do report outcomes, revenue uplift now accounts for over 15% of use cases with reported outcomes
- GenAI remains internally focused for now: 85% of genAI use cases target internal users
- Of the 15% of genAI deployments that are customer-facing, over half are concentrated in retail banking, where banks are enhancing chatbots, launching AI-powered avatars, and experimenting with agentic workflows that streamline the customer experience
- AI copilots are reaching scale: ten banks have now deployed genAI assistants to over 800,000 employees, though few have firm-wide productivity metrics
- Agentic AI is on the horizon: Nine banks have publicly disclosed agentic AI deployments, while the number of banking employees working on agentic AI has grown 13x YoY
- Three banks set the pace on agentic infrastructure: Capital One, J.P. Morgan and BNY have already detailed their agentic workflows – a prerequisite for scaling promising solutions amidst ongoing scrutiny from both internal and external stakeholders.
“Vertical genAI deployments often produce the best success stories because they’re embedded in defined processes like sales, onboarding, or fraud detection, making it easier to isolate their impact and quantify outcomes,” said Mousavizadeh in a statement. “But there’s a clear incentive for banks to up the ante and show stakeholders how their more advanced use cases – like customer-facing applications or agentic AI deployments – are delivering tangible value and first-mover advantage.”