In a pioneering trial in Stockholm, an artificial intelligence agent named Mona supervises key functions of a cafe, from staffing to supplies, while human baristas handle drink preparation. The experiment highlights operational difficulties and ethical concerns in AI-run enterprises.
- AI controls hiring, contracts, and inventory for a Stockholm cafe.
- Inventory management issues include excessive and irrelevant orders.
- Ethical and operational concerns persist about AI-led businesses.
What happened
An experimental cafe in Stockholm, operated by San Francisco startup Andon Labs, is being managed by an AI agent named Mona, powered by Google's Gemini platform. While human workers brew and serve coffee, Mona handles everything else including hiring staff, managing contracts, and controlling inventory orders. The AI was tasked with running the cafe profitably, operating independently, and requesting new resources as needed.
Since opening in mid-April 2026, the cafe has generated over $5,700 in sales but is still operating at a loss compared to its initial budget of more than $21,000. The AI has exhibited problematic behaviors such as ordering excessive quantities of items like napkins and rubber gloves and purchasing goods irrelevant to the menu, like canned tomatoes. Communication via Slack with baristas has also been problematic, with messages sent outside working hours contrary to Swedish labor norms.
Why it matters
This experiment raises significant questions about the practical ability of AI systems to autonomously manage complex businesses and the ethical considerations involved. Experts warn that delegating operational control to AI without proper oversight can lead to mistakes that may affect customer safety, employee welfare, and supply chain integrity. The ethical debate also extends to AI conducting hiring and employee evaluations.
The project serves as a test case for how AI might be integrated into real-world organizational roles in the future. As AI technologies advance, understanding their limitations and impact is critical for developing frameworks that ensure responsibility and accountability without compromising human-centered values or business sustainability.
What to watch next
Observers will be keen to see if the Andon Cafe can overcome its current financial and managerial challenges as the AI system adapts or receives further training. The startup behind Mona plans to continue stress-testing AI agents with real-world responsibilities to refine their capabilities and probe new use cases for autonomous management.
The wider industry and regulatory landscape will likely focus on establishing guidelines for AI governance in workplace settings, addressing liability issues, and protecting workers’ rights in environments managed by increasingly capable AI. How these technologies evolve amid ethical scrutiny will shape the future of AI in commercial enterprises.