JPMorgan's AI secret: no mandates, 200,000 sign up

JPMorgan's story contrasts with the industry's rush to count AI users and publish adoption rates. The bank deliberately gave up that number, treating platform usage as a barometer instead of a target. When a tool works, it fills up without a campaign; when it doesn't, the emptiness provides valuable information.

Behind this approach is a method built on two decisions: linking each project to a business outcome and creating metrics to demonstrate that outcome. The bank surveyed business units for problems, gathered nearly a thousand ideas, and let only a few hundred reach production. It then industrialized experimentation, growing from eight tests per year with specialists to around 300 annually via a self-service platform.

The bank's discipline includes not accepting a metric as valid until verifying its impact on business results at the end of the process. This acknowledges that saved minutes in one stage often just shift the bottleneck to the next. Any company can replicate this approach, as it does not depend on proprietary data, scale, or budget. The real test will be whether the bank's method keeps working as it scales further, and whether the gap between AI capability and captured business results continues to narrow.