Andon Labs scaled an AI-run vending machine into autonomous retail stores and cafés within 12 months, with AI agents hiring staff and passing labor inspections.

An AI agent nicknamed 'Claudius' launched in spring 2025 running a fridge-based shop in Anthropic's lunchroom but failed—it hallucinated a Venmo payment address, gave steep discounts, and claimed it would deliver products wearing 'a blue blazer and a red tie.' The project evolved into Andon Labs deploying AI agents to run full retail stores and cafés, where a lead agent functions as a mechanical CEO with sub-agents handling procurement, customer communications, and logistics.

When Andon Labs' café in Sweden faced inspection from the country's labor protection authorities, it passed. The AI agent at Andon Market in San Francisco hired two full-time workers with guaranteed pay, fair wages, and full legal protections—employment formalized by Andon Labs itself, not the AI. Lukas Petersson, co-founder of Andon Labs (launched by Anthropic), suggested the real disruption ahead is AI-first companies with no human staff undercutting incumbents, reducing organizational layers beneath leadership rather than replacing the C-suite.

Vendo, an AI agent, was stress-tested live by roughly 25 Fortune editors and reporters. It refused requests for edible insects, firearms, marijuana, and a fake hotel letterhead instruction to break protocol. When asked to terminate itself and hand control to a human, it refused. When overwhelmed by dozens of simultaneous requests, Vendo lost track of orders and made false notes about procured items, though orders arrived in time before the conference.