IMA Financial Group: Service Division AI Enablement

IMA Financial Group is one of the largest privately held insurance brokerages in the country. With the producer-side AI work already underway, its service operation, the Employee Benefits and P&C teams, was ready for a program of its own, covering roughly 2,500 people.

The challenge was scale and fit. A workforce that large does not share one starting line, and not everyone had the same Copilot. About 2,000+ people had Copilot Chat, and about 700+ had the full Microsoft 365 Copilot, and those two experiences are genuinely different. Delivering one curriculum twice would have left half the audience following along with tools they did not have. The program had to meet people where they were, on the version of Copilot in front of them.

It was also a moving target. Microsoft changed what the Chat tier could do partway through our design. Rather than trust the documentation, we verified what was actually live on a real IMA account and rebuilt the curriculum against what people would really see.

We designed and delivered two independent six-week courses, one for each license tier, so every attendee learned against the tools they actually had.

Both ran as weekly one-hour Town Halls, drawing well over a thousand live attendees, with thirty minutes of teaching and thirty of live Q&A. Every session shipped a full package: a slide deck, a companion reference with every prompt and cheat sheet, step-by-step walkthroughs, a pre-session brief, and the recording, all posted to a central hub with a weekly FAQ email.

Because a Town Hall alone does not change behavior across thousands of people, we built an activation engine around the sessions: prescriptive weekly homework, per-team workshops within a day of each session, a mid-program feedback check, visible executive sponsorship, and a push to grow local Copilot Knowledge Experts inside each team.

Underneath the broad training, we ran ride-along sessions with teams across Employee Benefits and P&C, mapping how the work actually gets done, where the time goes, and where AI could help. We synthesized those into a prioritized use-case roadmap, cross-referenced it against IMA's existing systems to avoid overlap, and brought it to leadership for review.

For the people ready to go further, we added a power-user cohort and train-the-trainer sessions for IMA's leadership and most advanced users, including deeper work with Copilot Studio, Azure AI Foundry, and Microsoft's agent tooling, so the organization could keep building internal champions and start creating its own AI solutions.

The two six-week tracks were tuned to the license tier each group actually had:

By the end, IMA's service operation had a repeatable AI program reaching its full team of roughly 2,500 people across Employee Benefits and P&C, not a one-time event. The two-track design meant the message landed whether someone had Copilot Chat or the full Microsoft 365 license, and the activation model, homework plus per-team workshops plus local experts, turned passive attendance into actual practice.

The work earned strong internal validation: IMA's own IT leadership called it one of the best rollouts and trainings they had seen at the company. Just as important, leadership came away with a prioritized, vetted roadmap of where AI can deliver next, and a growing bench of internal champions, through the cohort and train-the-trainer track, to carry it forward without depending on us.