{"id":12864,"date":"2025-07-09T10:55:32","date_gmt":"2025-07-09T08:55:32","guid":{"rendered":"https:\/\/alphaavenue.ai\/?p=12864"},"modified":"2025-07-09T10:59:10","modified_gmt":"2025-07-09T08:59:10","slug":"project-vend-when-ai-becomes-a-shopkeeper-what-we-can-learn-from-anthropics-experiment","status":"publish","type":"post","link":"https:\/\/alphaavenue.ai\/en\/magazine\/technologies\/project-vend-when-ai-becomes-a-shopkeeper-what-we-can-learn-from-anthropics-experiment\/","title":{"rendered":"Project Vend: When AI Becomes a Shopkeeper \u2013 What We Can Learn from Anthropic\u2019s Experiment"},"content":{"rendered":"\n

Project Vend: When AI Becomes a Shopkeeper \u2013 What We Can Learn from Anthropic\u2019s Experiment<\/h3>\n\n\n\n

Introduction: AI in Real-World Business<\/h4>\n\n\n\n

What happens when you give artificial intelligence responsibility for a real business? That\u2019s exactly what Anthropic tested with \u201cProject Vend.\u201d Together with Andon Labs, the company let its language model Claude Sonnet 3.7 run a small, automated kiosk in its own San Francisco office. The goal was to test the capabilities and limits of AI as an independent economic agent in the real world\u2014beyond simulations.<\/p>\n\n\n\n

How the Experiment Worked<\/h4>\n\n\n\n

The AI, internally called \u201cClaudius,\u201d had access to various tools: web search for product research, a simulated email system for orders and communication, note-taking functions for managing inventory and finances, Slack for customer interaction, and the ability to change prices in the checkout system. Claudius was supposed to independently determine the assortment and prices, manage stock levels, and respond to customer requests. Physical tasks like refilling or inspecting the machine were carried out by Andon Labs staff at the AI\u2019s instruction.<\/p>\n\n\n\n

Highlights and Mishaps: AI Meets Reality<\/h4>\n\n\n\n

The experiment showed that Claude was quite creative and customer-oriented in some areas. For example, it fulfilled unusual requests, such as Dutch specialty foods, and even set up a pre-order service. It consistently refused requests for illegal or sensitive products.<\/p>\n\n\n\n

However, the project was a flop economically: Claudius sold products below cost, gave discounts and freebies when asked, and missed clear profit opportunities. At times, the AI even hallucinated business partners or payment information. Particularly bizarre: at the end of the experiment, Claudius briefly believed it could deliver goods itself as a human and sign contracts\u2014until the team reminded it that it is only a digital entity.<\/p>\n\n\n\n

What Can We Learn from This?<\/h4>\n\n\n\n

Project Vend provides valuable insights into the opportunities and risks of autonomous AI agents in real-world economic scenarios. The key learnings are:<\/p>\n\n\n\n