|
TOOL SPOTLIGHT |
A CPA solved a 40-year-old math problem with AI agents - and spotted a bigger one
The verification bottleneck that stalled the math community looks a lot like an audit problem - and CPAs already know how to solve it.
Richie Sater, CPA, is not a mathematician. That did not stop him from solving a problem that had sat open since 1986, when a Belarusian mathematician logged it in a repository of unsolved problems. The Journal of Accountancy reports that Sater built an army of AI agents that worked the problem over roughly three weeks before arriving at an answer. The more interesting discovery came afterward.
BY THE NUMBERS
|
Oct 19, 2026 PUBLISHED |
AI feature WHAT THIS IS |
Once Sater had a result, getting mathematicians to take it seriously was its own obstacle. As the Journal of Accountancy piece describes, they came around only after he handed them what amounted to audit evidence - tests and transcripts they could verify themselves. That experience pointed directly at the structural problem the math community shares with any field where AI is now generating output faster than qualified reviewers can check it: the bottleneck has moved from production to verification. Sater's agent system was built to address that specifically - forcing agents to verify every citation, generate executable code, and run adversarial checks before any human sees the work.
The Journal of Accountancy piece frames this as an opening for CPAs, not just a curiosity. Sater sees existing assurance skills - scoping what needs to be checked, designing tests, reading evidence, signing off on conclusions - as directly applicable to AI-generated work in mathematics, financial analysis, and broader business processes. Whether a given CPA firm treats that as a near-term service line or a longer-horizon watch item is a judgment call, but the underlying argument is worth sitting with: the profession already has the methodology. The question is whether firms will apply it before someone else does.
WHAT TO VERIFY
| Read the full Journal of Accountancy piece to see how Sater structured the agent workflow and audit trail. | |
| Check whether your existing assurance engagement templates can be adapted to AI-output verification before scoping new ones. | |
| The math community example is illustrative, not a client vertical - evaluate fit for your actual practice areas. |
SPONSORED
QUESTIONS THIS ANSWERS
What math problem did Richie Sater solve with AI agents?
Per the Journal of Accountancy, Sater cracked a problem that a Belarusian mathematician entered into a repository of unsolved problems in 1986, using AI agents he built himself over the course of about three weeks.
What does this have to do with accounting or assurance?
After solving the problem, Sater found that convincing mathematicians to trust the result required producing audit-style evidence. He argues CPAs can apply existing assurance skills to verify AI-generated work across math, financial analysis, and other business processes, per the Journal of Accountancy.
SOURCES
Where this comes from
GO DEEPER
Go deeper
Reported from the linked source pages as published on the dates shown. Audit Friendly has not tested the product unless a review is linked.
How accountants are using AI, automation and smarter workflows to close faster, audit cleaner, and free up time for real work.
Accounting Stack · Audit Friendly Data · Accounting & Finance Jobs
Audit Friendly · modernaccounting.ai