I sit through a lot of panels on artificial intelligence (AI) in tax practice, and most of them stay at the strategy level: adoption curves, efficiency percentages, the usual vocabulary. My conversation with Ryan Corcoran of RSM and Kaceelyn Pouttu of Eide Bailly at the recent CPA.com + Blue J Tax Insights Forum went somewhere more useful. Two practitioners leading transformation at very different scales, one at a national firm and one at a large regional firm, shared what actually happens when you try to move a tax practice from compliance-only to advisory.
Their transformation journeys consistently aligned with the five pillars in CPA.com’s Tax Transformation Framework: strategy and governance, talent and culture, technology integration, service model, and operational excellence. Here's what stuck with me.
Strategy and governance: Someone has to hold the map
Kaceelyn didn't inherit a transformation title. She grew into the role because her firm's leadership recognized that she understood how the practice actually worked, and that mattered more than any technical AI credential. Her view is blunt: If you don't have a transformation team actively shaping your practice, the market will shape it for you.
Ryan's version of the same idea is more personal. He tells younger partners that this is a moment to set the direction of their practice group for the next decade, not manage business as usual. But he's just as clear about the failure mode. The number of possible AI use cases is close to limitless, and left unchecked, it becomes a problem. Someone has to own reining it in — deciding what gets built and consistently rolled out across the firm. Without that clarity, the firm is more likely to deal with chaos than tax transformation.
Talent and culture: The apprenticeship model didn't disappear, it changed shape
AI can short-circuit traditional staff development if firms let it happen by default. Handing a junior associate a finished AI draft doesn't build judgment if no one rethinks how that work gets assigned and reviewed. The firms getting this right are using AI intentionally, as a teaching tool, and deliberately redesigning how work moves through the practice so new staff still develop the skills they'll need.
The example that stuck with me most: A senior associate drafts a technical memo using AI. A junior staffer checks it for accuracy. Then it goes to a manager with an assignment to read every case cited and every revenue ruling, and come back two days later ready to talk through what the AI got right and where it missed something. Ryan calls this the old mentor-apprentice model, just compressed. The tool didn't replace the training rep, it changed what gets reviewed and how quickly people develop judgement.
Both Kaceelyn and Ryan agree that the key to recruiting and retaining people is helping staff feel like their work has value and counts for the impact to clients – helping them both to learn and use judgment along with the opportunity to exercise it regularly. Ryan pushes that same thinking further upstream into recruiting: A day of thirteen back-to-back interviews tells you almost nothing about how individuals can learn to apply judgment and work well with people. Firms that treat hiring as a longer relationship, rather than a single interview loop, end up retaining more of the people they bring in.
Technology integration: The expert steers the model, not the other way around
Ryan built a prompt for accounting methods work that reads a bonus plan, distinguishes fixed from non-fixed compensation, and surfaces the relevant tax regulations as it goes, teaching associates the citations along the way instead of handing them an answer. But he is equally direct about where AI goes wrong without that kind of steering. Ask a general question and the model can just as easily pull in an unrelated but statistically similar tax code section. People call that a hallucination. Ryan calls it an unsteered model doing exactly what it was asked, badly. Only a subject matter expert can define the boundaries it should search within.
Eide Bailly's numbers back up the payoff of getting this right. Kaceelyn notes that in the firm’s post-busy-season survey, 88% of staff named Blue J’s AI tax research tool as the firm’s most beneficial technology improvement, without being prompted to say so.
Service model: The billable hour is losing its grip
Neither firm frames the move away from hourly billing as a pricing exercise. It shows up as a measurement problem. Kaceelyn's firm is rethinking whether billable hours should stay the primary metric for junior staff. Ryan's angle is capacity: AI tax research and automated drafting are freeing partners from time-intensive tasks, creating space for business development and proactive planning that hourly incentives had quietly been squeezing out. If the metric doesn't change, the behavior won't either.
Operational excellence: The unglamorous work of not rushing it
My favorite story from the panel: RSM signed on for 4,000 Blue J AI research licenses in September and wanted the technology deployed firmwide within weeks. Their Tax Chief Operating Officer said no. The training materials weren't ready, governance hadn’t been established, and rolling out to an entire tax practice required a different level of preparation than a pilot program. It took roughly a year to get change management and the tax practice fully aligned. Ryan’s take now: The friction was the point, not an obstacle to it.
That experience reinforces one of the central messages in the Tax Transformation Framework: Successful transformation depends less on how quickly technology is deployed than on how deliberately people, processes and governance evolve alongside it.
The pattern underneath both firms
Strip away the difference in scale and RSM and Eide Bailly are running the same playbook: Find where the pain is sharpest, build a small proof point, bring in the right expertise to scale it, and measure the right things, not just hours. Neither firm waited for a finished strategy before moving. They're building as they go, deliberately, one validated use case at a time.
If you want the full framework behind these five pillars, our Tax Transformation guide walks through each one in more depth, along with practical steps for getting started.
Learn more about the CPA.com tax transformation framework
About the author
Brandon Allfrey, CPA, CGMA is the Senior Director of Tax Transformation at CPA.com where he helps CPA firms modernize tax compliance and expand advisory services by leading the strategy and development of innovative tax technology solutions.
