Some of the company’s most important work took a week by hand
Pricing a single incentive trip meant pulling rates from several long supplier proposals into one model, and it took the team a week. Every month, the operations team spent a week reconciling expenses and producing statements for the company’s advisors. Marketing spent hours preparing each content calendar, which left little room to plan ahead.
The interest in AI was already there. As one executive put it, “we know everybody’s using it,” and 76 percent of staff said AI could make their daily work easier. But only 44 percent knew which AI tools they could access, and most of the use that existed was quick chat questions and emails, not the work that took the most time.
Leadership set the rules first and asked for the push to come from the teams
Before any team sessions, leadership named an AI Leader with defined decision rights, approved two tools, and sorted data into three tiers, with client personal information, payment details and partner agreements kept out of AI tools entirely. It appointed a champion in each department and added AI measures to its weekly scorecard. It also set one test for every new use: does it elevate the work while preserving the personal connection that defines the company? Leadership wanted the use to come from the teams rather than be handed down.
Each team rebuilt its own process in Claude, with a team member at the keyboard
After the company-wide sessions, each department met for a pair of sessions. In the first, we mapped the team’s real workflows and chose the ones taking the most time. In the second, a team member built the tool live while I guided. Each build lives in a shared Claude workspace with the team’s instructions, examples and reference files, so anyone on the team gets the same result each time. The incentives team turned how it prices a trip into written instructions and a folder of past pricing models. Operations moved its statement and expense work from one person’s chat into a shared workspace. Marketing built a content calendar tool from its own planning process.
Work that took a week now takes hours
The incentives team now builds a trip pricing model in an hour instead of a week. Its time goes to checking the numbers and shaping the proposal for the client.
The operations team’s monthly expense reconciliation and advisor statements, a week of work every month, now take a few hours.
Marketing cut its content calendar prep from hours to 30 minutes, and the team now plans its content ahead rather than week to week.
Every department in the sessions rated them 4 or 5 out of 5, and by July company leaders were telling others that the push for AI was coming from the teams themselves.