Share with your CHRO
Voya Financial’s chief technology and operations officer Santhosh Keshavan makes the case that workforce trust, not technology capability, is the real constraint on enterprise AI adoption. The company pushed a five-hour generative AI literacy program to all 11,000 employees, shaped by more than 1,000 of them, and reached near-100% certification in three months. Today 98% of employees are actively using Microsoft Copilot, averaging 24 prompts per user weekly. A subsequent internal agent-building tournament produced nearly 300 employee-submitted agents competing for enterprise deployment.
What this means for your business
The companies most at risk of stalled AI programs aren’t the ones with bad technology choices. They’re the ones where the workforce quietly waits to be told what to think. Voya’s numbers are genuinely unusual: 98% active Copilot usage at a financial services firm with 11,000 employees isn’t a pilot stat, it’s a culture signal. If your own AI tools show strong procurement spend but thin daily engagement, the gap is almost certainly sitting in the trust layer, not the feature set.
The piece is authored by Voya’s own technology chief, so the argument naturally flatters a model where IT-led literacy programs drive adoption rather than, say, business unit autonomy or tool-first experimentation. That tilt is worth noting because it may understate how much the program’s success depended on specific conditions: a CEO-level mandate, a financial services culture already trained on compliance-heavy onboarding, and a workforce that couldn’t easily opt out. Five hours of training reaching 100% completion in three months doesn’t happen through good curriculum design alone. It happens when leadership makes non-participation uncomfortable. That’s a legitimate strategy, but it’s a different capability than “build good training.”
The agent tournament is the more interesting signal. When 300 employee-submitted agents emerge from a three-month-old literacy program, the organization has crossed from AI awareness into distributed AI judgment, meaning front-line workers identifying and prototyping process changes without waiting for a center of excellence to greenlight them. That’s the condition most CHROs and CIOs claim to want but rarely architect for. The question worth weighing now isn’t whether to run a similar program, but whether your current performance management and rewards structure would actually surface and promote the people who built the best agents, or route their work back into a backlog.
Based on reporting from AI transformation starts with building employee trust, originally published 2026-08-28 11:43:00.
