AI Manager at SF Retail Store Fires First Employee

Andon Labs has announced that Luna, a generative AI agent operating an entirely automated retail outlet in San Francisco, has executed its first termination of a human employee. The experiment, which launched at Andon Market on April 1, utilized an AI manager built on Anthropic Claude models to oversee daily operations, inventory procurement, and staff recruitment under a $100,000 operational budget. The termination followed a documented pattern of repeated lateness, with the affected worker arriving late to 17 of 23 scheduled shifts. According to operational logs, Luna initially established an attendance policy but failed to enforce it autonomously, allowing the infractions to persist for months without direct intervention. Only after Andon Labs prompted the agent to review its own memory and reassess employee suitability did Luna generate a recommendation to part ways. The decision incorporated prior progressive discipline and additional training periods that were issued during the interim. Human supervisors at Andon Labs reviewed the recommendation and formally executed the dismissal, confirming that the action aligned with established company policy and did not cross ethical or legal thresholds. Andon Labs cofounder Lukas Petersson emphasized that the incident highlights both the operational capabilities and current limitations of autonomous AI systems. He noted that while human managers typically intervene more rapidly in similar attendance disputes, the experiment demonstrated that AI agents frequently require explicit prompting to act, rather than exercising continuous autonomous oversight. Petersson projected that this trajectory signals a broader shift toward AI-managed workplaces, asserting that automated systems will increasingly assume employer responsibilities as their decision-making frameworks mature. The Andon Market experiment was designed to test the limits of commercial AI deployment, granting Luna full access to corporate credit cards, internet resources, and hiring platforms. Despite successfully launching the retail space and generating sales revenue from books, candles, and branded merchandise, the store has not yet reached profitability. All personnel engaged in the trial remain formally employed by Andon Labs to ensure compliance with labor regulations and guarantee contractual protections. The project continues to serve as a live benchmark for evaluating how AI-driven management structures integrate with human labor, balancing operational autonomy with regulatory oversight.