Oct 22, 2026 · 57 min
AI beats CPAs on close tasks, but workflow design decides the outcome
Frontier Models Outscore CPAs & What Clients Will Spend on Accounting
The episode tests whether accounting firms can capture AI’s speed and lower cost without losing context, judgment, or client trust.
- 1Frontier AI completed isolated month-end close tasks faster, more accurately, and more cheaply than 12 licensed CPAs.
- 2Approval-heavy workflows can erase AI’s efficiency, making exception-based oversight more valuable than reviewing every automated step.
- 3Accounting budgets rise sharply above roughly $1 million in revenue, while niche authority increasingly shapes AI-driven firm discovery.
Don't miss
The hosts unpack why CPAs working with Claude were often much slower than Claude operating independently, exposing approval-heavy review as the bottleneck.
The brief
Blake Oliver and David Leary examine a benchmark in which frontier AI completed realistic month-end close tasks faster, more accurately, and at dramatically lower cost than 12 CPAs.
The result comes with a boundary: isolated tasks cannot measure context, judgment, or client communication, and human review can turn an efficient system into a bottleneck.
The hosts argue that AI-native firms should automate clearly defined work and reserve people for exceptions, rather than forcing humans to approve every step.
Research based on nearly 2,000 prospective clients shows accounting budgets rise sharply after roughly $1 million in revenue, while relatively few businesses will spend $20,000 or more.
The episode closes on AI-driven firm discovery, where niche expertise and useful content may matter more than traditional location-based SEO or multiple local listings.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.