AI agent Mona struggles to run experimental cafe profitably

10 articles · Updated · The Associated Press · May 11

At Andon Café in Stockholm, the Google Gemini-powered system has made over $5,700 since mid-April, while less than $5,000 remains from a budget exceeding $21,000.
Andon Labs says Mona handles hiring, permits and inventory, but has overordered 6,000 napkins, 3,000 gloves and other unused supplies, while missed bakery deadlines have cut sandwiches from the menu.
The startup says the project tests how AI might run businesses, but researchers warn of ethical and accountability risks in letting systems manage staff, judge performance and make operational decisions.

This AI failed to run a cafe, so why do experts claim AI will soon manage trillions in global commerce?

Is the AI's chaotic management a failure, or a successful test revealing the true limits of autonomous business?

If an AI manager commits fraud, who is legally accountable: the code, the coder, or the company?