An agent can produce work very quickly: a report, a contract review, 1,000 lines of code. But producing work and checking work are two different jobs, and checking has not become faster.
A Worked Example
The video uses round numbers. An agent writes 100 reports in an hour. A careful reviewer can check five in that hour. By the end of the day, that pile of unchecked work is the real risk. It looks finished. It reads well. And some of it is wrong.
Subtracting, 95 of the 100 are still unchecked after one hour. That subtraction is this publication’s arithmetic on the video’s example.
The gap between making and checking is where the risk piles up. Round numbers for illustration.
The Audit Problem
This is the audit problem. When creating work becomes cheap, verifying it becomes the bottleneck. The video gives four ways to respond.
Four Ways to Respond
Keep a record of every step the agent took, so the work can be traced. Build automatic checks for whatever a machine can test. Review a small sample chosen at random. And save human attention for the cases where an error costs the most.
What You Are Buying
The problem also changes what you are buying. An agent that does the work but leaves no evidence hands you the audit. An agent that shows its sources and its steps has done part of the checking for you.
The checker need not always be a person. A second system can review the first. But then someone has to check the checker. The question never fully disappears. It only moves.
The AI Audit Problem builds on AI Agents. An agent writes 100 reports in an hour while a reviewer checks five, so the audit problem starts with agents.
Human in the Loop vs Human on the Loop builds on The AI Audit Problem. On the loop, a person reviews only flagged or sampled work, one of the audit lesson’s ways to respond to cheap production.
The AI Audit Problem builds on Process Map(lesson unlocks soon). When producing gets fast, checking becomes the step where work waits, the slow step a process map is drawn to find.
The AI Audit Problem builds on Verification Cost(lesson unlocks soon). The audit problem is verification cost at scale: making work got cheap, checking it did not.
The Business Engineer framework maps where each part of the AI stack sits and who holds the leverage. These pills are short lessons from the same library.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
