Artificial Intelligence
Written by: CDO Magazine
Updated 12:40 PM EST, November 16, 2026
Making generative AI available to employees is one step. Proving it changed anything in a way finance accepts is another.
In the first part of a three-part CDO Magazine series, Richard Leaton, VP, Data, Analytics, and AI at Novelis, speaks with Bob West, Data and AI Security Officer at Cyera, about measuring AI adoption honestly and preparing for autonomous AI.
For Leaton, the number of AI licenses a company buys only measures procurement. Left unused, those licenses are “shelfware with an invoice.” Instead, the executive tracks monthly active users and the kinds of work done in the tool.
Drawing on six industry studies, Leaton built a transfer function that converts usage into hours and dollars reinvested, tracked weekly on a 180-day rolling basis. Correcting an email earns few hours in that model, while creating a PowerPoint earns more and building an agent that does the work earns more still.
Leaton concedes the metric may be squishy because productivity gains are hard to tie to revenue or cost savings. Transparency about what it shows has kept finance comfortable.
With agents, no human checks the data one last time. Leaton puts data readiness and guardrails first, then names process as the bottleneck to fix, since unrationalized processes make agents overspend.
Highlights from the conversation:
On the move? Take this conversation with you. If you don’t have time to finish the video, you can stream the complete, uninterrupted audio edition of this episode on the go.