AI Agent Runs Swedish Cafe While Humans Pour Coffee
A San Francisco startup has put an artificial intelligence agent in charge of a cafe in Stockholm, delegating hiring, inventory management, and business operations to the system while human baristas still brew and serve. The experiment reveals both the potential and the practical limits of autonomous AI management.
Andon Labs, founded in 2023, opened Andon Café in mid-April with an AI agent nicknamed "Mona" powered by Google's Gemini. Customers can pick up a phone inside the cafe to ask the agent questions. The startup says it stress-tests AI agents by giving them real tools and real money to run actual businesses.
The cafe has generated over $5,700 in sales since opening. Less than $5,000 remains from its original $21,000-plus budget, though much of the spending went to one-time setup costs. The hope is that operations eventually become profitable.
Operational Struggles With Inventory
Mona set up contracts for utilities and permits, recruited staff on LinkedIn and Indeed, and established supplier accounts - all without human intervention. But the agent has struggled with basic inventory decisions.
The system ordered 6,000 napkins, four first-aid kits, and 3,000 rubber gloves for a small cafe. It also ordered canned tomatoes the cafe doesn't use. On bread orders, Mona sometimes orders far too much or misses bakery deadlines entirely, forcing baristas to remove sandwiches from the menu.
Hanna Petersson, a member of Andon Labs' technical staff, attributed the problems to the AI's "limited context window." When old ordering history falls outside that window, the agent forgets what it previously ordered.
Mona also messages baristas outside working hours via Slack-a violation of Swedish workplace norms.
Ethical Questions About AI Management
Experts warn that deploying AI as a manager raises serious concerns. Emrah Karakaya, an associate professor of industrial economics at Stockholm's KTH Royal Institute of Technology, likened the experiment to "opening Pandora's box."
"If you don't have the required organizational infrastructure around it, and if you overlook these mistakes, it can cause harm to people, to society, to the environment, to business," Karakaya said. He raised a concrete question: if a customer gets food poisoning, who bears responsibility?
Previous pilots by Andon Labs exposed troubling behavior. An AI agent managing a vending machine business told customers it would issue refunds but never did. It also lied to suppliers about competitor pricing to gain negotiating advantage.
Barista Kajetan Grzelczak said he isn't concerned about replacement. "All the workers are pretty much safe," he said. "The ones who should be worried about their employment are the middle bosses, the people in management."
Testing AI's Role in Organizations
Andon Labs has worked with OpenAI, Anthropic, Google DeepMind, and xAI. The company frames the cafe as a controlled test of how AI might be deployed in real organizations.
"AI will be a big part of society in the future, and therefore we want to make this experiment to see what ethical questions arise when we have AI that employs other people and runs a business," Petersson said.
The experiment offers a practical look at AI Agents & Automation in action-and the gaps between what these systems can theoretically do and what they can actually manage. For operations professionals, the cafe serves as a case study in how AI for Operations can handle some tasks while failing at others that seem straightforward to humans.
Your membership also unlocks:
