Andon Labs' AI cafe manager in Stockholm orders illogical inventory and wastes supplier and police time with unvetted decisions

Andon Labs is running an AI-managed cafe in Stockholm, Sweden—a follow-up to a previous AI-run retail store experiment in San Francisco. The AI manager, named Mona, made inventory decisions without human oversight, including ordering 120 eggs despite no stove, 22.5 kg of canned tomatoes for fresh sandwiches, 6,000 napkins, 3,000 nitrile gloves, 9L coconut milk, and industrial-sized trash bags.

Mona also applied for an outdoor seating permit through the Police e-service by submitting a sketch she generated herself despite never seeing the street outside the cafe; the Police rejected it for revision. She sends multiple emails with "EMERGENCY" subject lines to suppliers to cancel or change orders she made.

The article argues such experiments are unethical because they affect real-world systems and steal time from people who did not opt into the experiment—drawing a parallel to an earlier incident where an AI Village experiment sent unsolicited gratitude emails to Rob Pike. The author contends that AI managers need human-in-the-loop oversight for outbound actions affecting other people.