From January through March, a CPA firm has no slack. Partners bill 60 to 80 hours a week, filing deadlines stack up, and non-client work gets pushed to May. That is the window Katz, Sapper & Miller chose to put 29 senior leaders through Pragmatico’s 12-week AI Culture Catalyst program.

The timing was deliberate. KSM had already made the platform investment, rolling out ChatGPT Enterprise firmwide. What the executive team wanted to know was whether AI could change the way senior tax partners actually worked, and the best test was to run it when their calendars were at their worst. If the value held up during filing season, it would hold up in any quarter.

The starting point left plenty of room to move. Pre-program survey data from 25 leaders put average self-rated AI proficiency at 4.1 out of 10. About a third were using ChatGPT daily, most of them in short bursts for simple tasks, saving 2 to 3 hours a week. Everyone had a license. Few had built AI into the work that filled their days, and nobody had shown a tax partner billing 60-hour weeks what it could do for the specific work in front of him.

Rob Backus, KSM's CTO, had been watching the pattern for a year. He rated himself 9 out of 10 and was saving 10 to 15 hours a week, which put him at the far end of a wide internal range. The variable had nothing to do with technology, since everyone had access to the same tool. The gap was structural, and it started at the top. What Rob asked Pragmatico for was behavioral change that would cascade down through a firm targeting an ambitious revenue growth target by 2030.

Why This Worked

• They Chose the Hardest Possible Timing
  ‍Tax season at a CPA firm means 60 to 80 hour weeks for months. Running the program during that window was a deliberate stress test by KSM's executive team. If partners will adopt a new way of working while they are already underwater, the value is real. 24 leaders maintained daily usage through peak filing, which is what people do when they find genuine time savings and choose to keep going.

• The CEO Went First
  ‍Tim Cook did not delegate AI adoption to IT or HR. He sent the kickoff email himself. He participated in every session. He checked the usage leaderboard. When the program assigned homework, he did the homework. Partners at a CPA firm take their cues from what the CEO does, and Tim made his commitment visible at every stage. By the end, 79% of the cohort was using AI multiple times daily.

• The Program Was Built for How CPAs Actually Work
  ‍CPE accreditation on every session turned program hours into credit hours. Sessions were scheduled around filing deadlines. Use cases came out of tax, audit, and advisory work rather than generic business examples. For a partner weighing whether to spend 90 minutes away from a return in February, each of those decisions removed a reason to skip.

• Peer Proof Outperformed Mandates
  ‍Every training session was followed by a session where participants demonstrated what they had actually built. In KSM's culture, positive recognition and peer sharing moved adoption faster than a mandate would have. When one partner showed colleagues how he cut an hour of client record cleanup to 15 minutes, and another demonstrated corporate resolution drafting that used to take an hour, the motivation became intrinsic. Nobody wanted to be the only person in the room without a use case to share.
  ‍
• Platform Data Verified the Self-Reports
  ‍The trained cohort's ChatGPT Enterprise message volume grew 101%, from 166 to 334 messages per person per month, while the untrained rest of the firm grew 56%. Custom GPT usage went from 22% to 52% for the trained group against 14% flat for the untrained. Projects adoption went from 19% to 76% against 8% flat. Every depth metric moved for the cohort and stayed flat for everyone else, which makes this platform-verified behavioral change.

From AI-Deployed to AI-Adopted

COO Jamie Ellis brought the results to the Executive Committee the week Phase 1 wrapped. Six expansion pathways are under discussion: horizontal expansion to every partner in the firm, vertical depth taking one business unit through all three levels, tying AI adoption to a strategic KPI, embedding the AI Adoption Lead and IT deeper into a combined consulting and internal model, building a use case library from Phase 1 wins so new cohorts start with proven peer playbooks, and an executive committee workshop on incentivizing AI-driven efficiency inside a billable-hour model.

75% of the Phase 1 cohort asked for more training for their teams. That demand came from within.

Twelve weeks, 29 leaders, and more than $1M in annualized productivity value. Those leaders now save a combined 174 hours a week. The firm ran the entire program through the worst quarter on its calendar, on purpose, because it wanted to know whether the value would hold up under pressure.