AI Workshop For A Charlotte Insurance Agency | Alyra

A half-day AI workshop for a 16-person insurance agency in Charlotte, NC. We built a submission checker live in the room and put it on all 16 seats that day.

From Training To Application: An AI Workshop For An Insurance Brokerage

AI Workshops. CAP Insurance Group. Insurance. Independent agency in Charlotte, North Carolina.

At a glance

CAP wanted the whole agency on Claude, working the same way with it and sharing what worked instead of each person discovering it alone. The immediate aim was a standard: one agreed approach to the work, and a clear view of which tasks were worth automating first. The longer one was a team that keeps spotting the next workflow worth automating, without waiting to be shown where to look.

Role-based training took all sixteen people onto Claude in a single afternoon, sales included, with a session on faster lead follow-up. Adoption reached 100% of the agency, with each role working on their own tasks instead of watching a demo. A commercial submission checker, built live against CAP's own forms, went out to every seat the same day. Six weeks on the team is building its own projects per account and running an email automation nobody specified.

The hard part was not buying the tool

Claude was already in the building, and about a third of the team was using it, mostly for lookups.

CAP Insurance Group is a fast-growing independent agency in Charlotte, writing both personal and commercial lines and preparing for its next stage of growth.

Leadership had already made the call on AI. Claude was bought and deployed before most agencies their size had started the conversation. That is the harder decision, and CAP had made it.

What came next is the part nobody warns you about. The tools change faster than a team can teach itself to use them. The people seeing real returns had worked it out on their own, and what they learned stayed with them.

Each side of the business was also working from its own knowledge. Personal lines, commercial and sales answered the same question from scratch, in different ways, so something solved in one team rarely reached the next. That landed on Charlie and Andy, who were selling and servicing their own books while running the agency. The principals wanted that knowledge out of their own heads and into a system.

So they came to us, and they were clear about what they did not want. Not a training session. They wanted the whole team brought to the same place at the same time, every person faster inside their own workflows, colleagues sharing what was working, Claude as the first move on a hard task, and guardrails in place before anything reached a client. The point was never this version of the tool. It was leaving people confident enough to keep up with the next one.

Learn how the agency actually works

Nothing was designed until we understood how the agency runs. A two-hour discovery session with leadership, held before the workshop, mapped the teams, the systems, the volumes, and the places where work already broke down.

That produced a written summary, delivered before the workshop, so the agenda could be built on CAP's own constraints and goals. Charlie and Andy, the agency's principals, approved it before the workshop began.

Decide what to build first

The workshop was deliberately kept at a basic to intermediate level. The objective was cultural before it was technical, and keeping it there gave everyone the chance to learn and to see where Claude fits their own work. Candidate work was organized into three buckets leadership could argue about: defend revenue, cut errors and omissions exposure, and give the team time back. Every idea was then placed on a grid of quick wins, big bets, fill-ins and not now, so everyone could see what was worth doing first and what could wait.

The producers got their own pass, and faster lead follow-up came out of it and went onto the list with everything else. The grid also took things off the table. One legacy system went to not now, and the room spent the rest of the session on work that could ship that day.

Build the top-ranked workflow live

Commercial submission intake came first in the team's own ranking, and we built it there and then, in the room, with the people who would use it. It was built against CAP's own paperwork, not a template. The agency supplied its own forms and reference documents the day before. The result tells a producer whether a submission is ready to go out, before it reaches an underwriter. It was deployed the same day at the organization level, so all sixteen seats had it that afternoon. Nobody had to rebuild it from memory.

The session also put a data foundation underneath it. The agency's knowledge had been spread across different places, so we set up one source of truth in SharePoint, holding the forms and reference documents everything else would read from. Everything built on it reads the same documents every time. A written AI Use Policy shipped alongside it, setting the rules for anything client-facing before there was anything to govern.

Everyone is on Claude, and nobody stopped there

The submission checker is the narrower question. It went to every seat the same day, and six weeks later the clearer result is that adoption of the method outran adoption of any single tool: the team can now build its own skills and automations without waiting on anyone to build them for it.

What the checker did change is the standard. A submission now carries what it needs before it goes out, so it reaches the account manager quoting it ready to work.

The adoption that did take hold went somewhere nobody specified. Custom skills were built for CAP's own staff during the workshop, and the team carried the method on from there: people are building their own Claude projects and sharing them against specific accounts, and running an email automation that gives them a weekly roundup. None of it was on the agenda. It came from people who had been shown how and went looking for the next thing themselves.

What happens next is being decided inside CAP. They are running a roundup with the sales team to work out what to automate, with one-to-one sessions with the producers after it to build out their own automations and skills. The staff reach for Claude on their own work now in a way they did not before the session.

"After our workshop with Alyra, our team has seen a significant increase in usage with Claude, including building new skills, projects, and automation. It’s really helped our team lean in and become AI forward in their thinking and efficiency." Charlie Sternberg, Principal, CAP Insurance Group

Sixteen staff, one half-day session, a submission checker built live in the room, and a team that kept building its own projects and automations for weeks afterwards.

Start a conversation