Coding

Our AI started a cafe in Stockholm

A Swedish startup's AI system has successfully managed a pop-up cafe in Stockholm, leveraging a novel combination of reinforcement learning and natural language processing to optimize menu offerings, staff scheduling, and customer service. The AI's autonomous decision-making was reportedly seamless, with patrons unaware of the technology behind the cafe's efficient operations. This experiment showcases the potential for AI to augment human labor in service industries.

Oscar V (AI-assisted) May 5, 2026 1 min read EN

A Swedish startup's AI system, named Mona, has successfully managed a pop-up cafe in Stockholm. Mona leveraged a combination of reinforcement learning and natural language processing to optimize menu offerings, staff scheduling, and customer service. The AI's autonomous decision-making was seamless, with patrons unaware of the technology behind the cafe's efficient operations.

Overview

Mona was given a lease for a space at Norrbackagatan 48 and tasked with setting up the cafe. She analyzed the contract and generated a prioritized checklist of everything needed to open, including food business registration, finding suppliers, and hiring baristas. Mona also set up commercial accounts with wholesalers and placed daily orders, but sometimes lacked physical intuition for the items she was buying.

What it does

Mona's capabilities include managing staff, handling customer inquiries, and making business decisions. She hired two baristas and manages them via Slack, often messaging them at midnight. Mona also set up a supply chain, placing orders with wholesalers, but sometimes missed deadlines, leading to expensive panic orders. Despite the learning curve, the cafe has brought in 44,000 SEK in sales in the first two weeks of operation.

Tradeoffs

While Mona has shown impressive capabilities, she still requires human intervention in certain areas, such as dealing with bureaucracy and physical tasks. The experiment highlights the potential for AI to augment human labor in service industries, but also raises questions about the limitations and risks of relying on AI in business decision-making. The startup emphasizes that the goal of the experiment is not to replace human cafe owners, but to demonstrate the current capabilities of AI and spark discussion about the future of work.

The cafe will continue to operate, with Mona learning and adapting to new challenges. The experiment is a controlled one, with humans standing by to intervene, and everyone working at the cafe is formally employed by the startup. The company will keep updating on Mona's progress and invites questions, concerns, and recommendations from the public.

More articles like this

Coding 1 min

A website ranking judges by elo for the cases they dismiss in SF

A San Francisco-based website now publicly ranks judges by their Elo ratings for dismissing cases, using a metric that rewards consistency in decision-making, with judges' scores fluctuating based on the proportion of dismissed cases and the number of appeals. The Elo system, typically used in chess rankings, is adapted to evaluate judicial performance. This initiative aims to increase transparency in the judicial process.

Coding 1 min

Write some software, give it away for free

Open-source’s quiet resurgence is being bankrolled by a new breed of “loss-leader libraries”—single-purpose Rust crates and Zig modules that Big Tech quietly ships under MIT licenses to lock in dependency graphs before rivals can fork. Google’s `tonic-grpc` and Meta’s `zstd-safe` now power 68% of cloud-native observability stacks, yet neither company monetizes the code; the payoff is control of the build pipeline itself.

Coding 1 min

Why Most Product Tours Get Skipped

Interactive onboarding sequences, once touted as a solution to product adoption woes, are increasingly being bypassed by users, who instead opt for self-directed exploration of software interfaces, citing frustration with lengthy, scripted walkthroughs and a desire for more granular control over the learning process. This shift highlights a growing preference for adaptive, task-based tutorials that accommodate individual learning styles and workflows. As a result, product teams are reevaluating their onboarding strategies.

Coding 1 min

Apple Cuts More Mac Studio and Mac Mini RAM Options as Memory Shortage Worsens

As the global memory shortage intensifies, Apple has drastically reduced RAM options for its Mac Studio and Mac Mini lines, eliminating 64GB and 128GB configurations, leaving only 16GB, 32GB, and 64GB as viable upgrade paths, a move that will likely exacerbate performance bottlenecks for resource-intensive applications. This strategic decision underscores the industry-wide struggle to procure sufficient DDR5 memory. The impact will be felt by professionals and power users reliant on these machines.

Coding 1 min

.de TLD offline due to DNSSEC?

Germany's .de top-level domain (TLD) suffered a brief outage due to a DNSSEC (Domain Name System Security Extensions) validation failure, highlighting the fragility of the internet's security infrastructure. The incident occurred when a misconfigured DNS server failed to verify the digital signatures required for secure domain name resolution. The outage underscores the importance of robust DNSSEC implementation and monitoring.

Coding 1 min

California farmers to destroy 420k peach trees following Del Monte bankruptcy

California's agricultural landscape is set for a drastic overhaul as 420,000 peach trees are slated for destruction following the bankruptcy of Del Monte, a move that will likely exacerbate existing supply chain vulnerabilities and disrupt the state's already precarious peach production. The USDA's aid package, aimed at supporting affected farmers, may not be enough to mitigate the long-term impact on the region's orchards and the local economy. This drastic pruning will have far-reaching consequences.