U.S. Bank Chief AI Officer Prashant Mehrotra set out six steps for moving AI work from trials to measured value. He spoke at American Banker's 2026 Digital Banking Conference. The steps are picking the right work, defining outcomes, productising and reusing, building skills, scaling responsibly and preparing for agents. Mehrotra said the bank targets high-frequency, repeatable work rather than one-offs. He said value must be measured at every gate, from pilot through to enterprise scale. The bank is building reusable components that teams assemble rather than rebuild. It has rolled out role-specific AI training to its whole workforce. American Banker maps the six steps to its own grid of measurement, scaling and governance.

• 70,000 employees in persona-based training
• 6 actions in the playbook