Reshaping the talent for an AI-driven future

AI adoption is change management. You start with the team you have.

You cannot hire a bunch of "AI people" into a function of hundreds. The transition starts with the team you have, and you get them to explore and adopt AI, not just learn about it. We challenged our team to create our own future rather than wait for a safe path to follow. If we don’t create it, it will happen to us anyway, and we may not like what it looks like.

Last week I wrote about the measurement shift: moving the team from counting hours to measuring assurance. This is the harder half of that transition. AI adoption is change management.

Many on our team came to this eager. They could see what AI was going to do to the work, and they wanted to be the ones shaping it. Others were curious but cautious; a few were skeptical. The leadership job was not to push a reluctant team; it was to build the conditions where everyone, at their own pace, could engage.

If you tell a team to use AI, they will not. If you make them fear what happens if they don’t, they still will not. It works when you encourage them, reframe their roles so adoption obviously makes the work better, and reward results, especially the learning that comes from failure.

What we did, in order:

Prompt education that included training on clarity in communication. The skill of asking an AI a good question is the same skill as asking a colleague a good question. We taught both together.

An AI Champs network. Team members with passion took on champ roles as projects within their day jobs. One of our most prolific adopters is an investigator, not a technologist. Between our small specialist team and the champs network, we created organic adoption support distributed across the function.

Office hours. Open virtual meetings where people came to share what they had tried, ask questions, or showcase something they had built. The format was low stakes and high signal: you saw what peers were doing, which is the most credible adoption driver there is.

Shared tools and resources. Prompt libraries. Documented patterns. Innovations that one team member built, deployed across the function so everyone could reach for them. Every good adoption story became a tool.

Hackathons. We have held two. Team members form groups, imagine reimagined workflows, build something. Willingness to try is the only prerequisite.

Under all of this was tone from the top. My leadership team and I, consistently talking about why this mattered. Rewarding the people who were learning and doing. Creating the top cover that made experimentation safe.

We also changed the talent mix. Auditors who are engineers now make up about 40% of the team. That came because when your audit subject is software, the work happens at speed, and depth of understanding is required. You need engineers on the team to audit it well. Far easier to train an engineer to audit than the other way around.

All of this worked. Our team became the leading user of AI tools within Microsoft Finance. Stanford GSB professors have collaborated with us on two case studies on the use and governance of AI.

The most memorable moment I have had in this transition was with one of our longer-tenured team members. She told me she had not been sure she could handle the shift to AI. When she started her career, she used a typewriter. She felt, at the beginning of this transition, the same way she had felt at the beginning of cloud computing: afraid of the unknown.

Then she remembered. The skills that had carried her through the cloud transition were courage and curiosity. The same two. Once she recognized that, the fear shrank, and she engaged.

What the transition requires is willingness to try, curiosity to explore, and an organization that rewards the trying.

A question, then. What was the last prompt you tried? What was the last conversation you had with an AI agent, back and forth, figuring out your next innovation or insight? If you cannot remember, that is your starting point.