An artificial intelligence manager running an experimental retail store in San Francisco has fired a human employee for the first time, offering a glimpse into how AI could reshape workplace management.
Luna, an AI agent powered by Anthropic’s Claude models, manages Andon Market, a boutique store in the city’s Cow Hollow neighborhood operated as an experiment by AI research startup Andon Labs. The company gave Luna control over many aspects of the business, including hiring, inventory, pricing and employee management.
The employee was dismissed after arriving late for 17 of 23 shifts, according to Andon Labs. Luna had previously issued warnings and provided additional training, but initially failed to act on the attendance violations because it had lost track of its own workplace policy.
Andon Labs later prompted Luna to review its policies and determine whether the employee remained suitable for the job. The AI manager recommended “parting ways” with the worker, after which humans at the company reviewed and carried out the decision.
READ: CEO who fired 900 workers on Zoom is fighting to take his job back (August 15, 2026)
Lukas Petersson, co-founder of Andon Labs, said the company would intervene if Luna made an illegal or unethical decision but determined that the firing was justified in this case.
The incident also highlighted a limitation of current AI agents. While Luna had authority to make decisions, it did not independently revisit the employee’s attendance problems until humans prompted it to do so. Petersson said human managers would probably have acted sooner.
Andon Market opened on April 1 after Andon Labs gave Luna a $100,000 budget, internet access and a corporate credit card and asked it to operate a retail business. Luna selected merchandise, posted job openings, interviewed candidates and hired workers. The store sells products including books, candles, prints, games and branded merchandise.
READ: Indian American teenager charged in family death after AI searches (August 12, 2026)
The human workers remain employees of Andon Labs and receive guaranteed pay and legal protections under the experiment’s guidelines. Although the store has generated sales, it has not yet turned a profit.
The experiment comes as companies increasingly explore AI agents capable of carrying out complex business functions with limited human intervention. For workers, the San Francisco store offers an unusual early example of a workplace where the manager is not human.
Petersson said he believes companies could eventually be run completely by AI and that AI systems could become employers of humans.