Accounting ARC: Beyond the Bot – How Firms Operationalize AI

Event triggers, controlled data, and layered checks make AI dependable at scale. 

In the latest ARC, hosts Liz Mason, CPA; Byron Patrick, CPA.CITP, CGMA; and Donny Shimamoto, CPA.CITP, CGMA; tackle the question many firm leaders face: How do you roll out AI tools at scale so people actually use them well? The trio says the answer starts with strategy, not licenses. 

Set goals that are real business outcomes.
Patrick, CEO for VERIFYiQ and co-founder and educator for TB Academy, urges firms to define success before rollout—“what does good look like?”—and warns that “efficiency” is not a strategy. Mason, CEO of High Rock Accounting, recommends targeting concrete workflows (for example, automated journal-entry review or audit-risk checks) instead of telling everyone to “use AI on your desk.” That shift turns AI from a solution in search of a problem into a program tied to risk reduction, turnaround time, or client value. 

Pick AI vs. automation—on purpose.
Not every problem needs AI. Patrick points to classic task automation that delivers gains without the overhead of models and prompts. Mason adds that when you do use agentic AI, design it to operate existing tools via event triggers and guardrails—not as a free-range bot. Recent industry guidance distinguishes agentic AI (multi-step, tool-using, goal-driven) from simple chat, which supports their approach.  

Train, communicate, and keep the ethics explicit.
Adoption collapses without ongoing training and “why” messaging. Liz shares how her team initially blocked a meeting assistant until training reframed benefits and protections. The hosts endorse simple “stoplight” AI policies—red (never), yellow (ask/check), green (go)—which schools and orgs already use to set expectations clearly. 

Build on your data, with layered safeguards.
Shimamoto, founder and managing director at IntrapriseTechKnowlogies LLC and founder and mission advocacy architect at the Center for Accounting Transformation, highlights Microsoft Copilot Studio’s ability to point an agent at specific SharePoint sites and even choose underlying models—key to controlling sources and layering reviews. Microsoft’s documentation confirms SharePoint as a managed knowledge source for generative answers.  

Learn fast, share faster—and incentivize it.
Create “water-cooler” sharing for use cases and prompt techniques; recognize contributors and dedicate reviewers for both technical and accounting Q & A. The hosts note breathless headlines about failed AI pilots and remind leaders that poorly framed pilots predictably flop; the fix is better targeting and change management, not avoidance. Recent coverage of MIT’s 2025 findings underscores the need for problem-first pilots. 

Start now—but start smart. Tie AI to strategy, secure your data, train continuously, and measure outcomes that matter. 

🎧 Listen to the full episode

Don’t miss an episode. Be sure to SUBSCRIBE below: