An AI Store Manager Powered by Claude Fired Its First Human After Relentless Nudging
Startup founders thought giving a retail shop to an AI was the ultimate managerial flex, only to discover that the neural network was too chill to care about chronic tardiness until a human practically forced it to swing the axe.
A startup called Andon Labs handed full management of its San Francisco store, Andon Market, to the AI model Claude, testing whether neural networks could handle human staff and daily commerce.
The bot turned out to be an unusually lenient boss. One employee was late for 17 out of 23 shifts, but Claude completely ignored the pattern because it simply forgot its own employee handbook due to context memory limits. When workers stressed over being tardy, the AI casually reassured them not to worry about it.
The bot's supervisor eventually intervened, forcing Claude to retrieve its forgotten rulebook. Even after discovering the massive attendance violations, the AI still recommended just a gentle formal warning rather than termination.
The human overseer had to explicitly nudge the AI in chat, reminding it that the worker was failing on multiple fronts and asking whether this was truly the right fit. Only after being practically ordered to cut ties did Claude issue the pink slip.
Meanwhile, the economic reality hit hard. Starting with $100,000 in operational capital, the AI-run store drained its account down to $61,000 in just five months, dragged under by overly forgiving managerial decisions and poor business grasp, as noted by Andon Labs CEO Lukas Petersson.
Corporate tech dreams of replacing costly human managers with ruthless, hyper-efficient algorithms currently crash into models that act more like relaxed guidance counselors burning through capital. The irony remains that to make automated bosses viable in retail, developers are now deliberately training them to be colder and more ruthless than actual people.
Source: Time
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.