AI enablement across two very different audiences

Two populations, one problem

Assurant handed everyone a Copilot license. Licenses do not teach anyone anything, so most people opened it once and went back to what they were doing.

At the other end, a handful of leaders and analysts were ready to go past chat and build things. Nothing existed to get them started. Whoever figured it out did so alone, and what they learned stayed with them.

Both groups needed teaching. They needed different teaching.

AI Pit Stop

I write a monthly newsletter for the Copilot population. It goes to business people with no interest in becoming technical and no time to try.

Each issue runs use cases and how-tos on the tool they already have. I include advanced prompting for anyone who wants to level up, and a section on the art of the possible to show people what is worth trying next. I record an audio version so people can listen while they drive or walk.

Around 150 people read each issue. I do not have Copilot usage data from the platform admin, so I track the newsletter by readership and by what comes back to me: likes, thank-you notes, and people asking how to do the thing they read about.

The Claude Code Onboarding Experience

For people going past Copilot, I built an onboarding experience that assumes you have never opened a terminal. It covers Windows and Mac, walks a new user through the basics of the Claude Code CLI, and teaches the handful of slash commands that make the tool usable on day one.

You point Claude at the folder and start working. Within a session someone can pull apart a dataset or build an HTML report, the work they would otherwise queue behind an analyst.

The part I cared most about was the first ten minutes. A blank terminal tells most business users this tool was built for engineers, and they close it. The onboarding gets them past that before they decide.

Showing the work

Our team runs a weekly brown-bag where we demo what we have built with Claude Code and then get people using it in the room. I present my own work there and help attendees get started on theirs.

The sessions moved the numbers. Global Automotive leadership now sits among the heaviest Claude Code users at Assurant, at the executive level and one layer down. Credit spend measures consumption rather than value, but it tells you these leaders work in the tool daily instead of approving budget for something they have never opened.

One senior leader has gone further and builds local applications and prototypes on his own. I work with him on best practices and on lining his work up with where we want development to go.

Ongoing

I publish AI Pit Stop every month. The tool landscape turns over faster than that, so what I recommend changes between issues.