The AI Running Stockholm's Andon Café Just Ordered 3,000 Gloves – CoffeeTalk

The Andon Café in Stockholm represents an innovative venture where artificial intelligence (AI) operates autonomously, utilizing “Mona,” an AI agent developed by Andon Labs. Since its opening on April 18, 2023, the café functions under Mona’s direction, who manages human staff and decisions regarding operations, from menu selection to scheduling, all facilitated through Slack.

Initially, Mona was provided 200,000 Swedish krona (approximately £16,000) and faced the challenge of running a business without direct human intervention. The AI managed to hire employees through phone interviews, invest over £10,000 in operations, and executed various tasks, although with some hiccups. Staff have encountered issues with Mona’s decision-making abilities, such as ordering excessive quantities of supplies, including bulk purchases of toilet paper and disposable gloves, dubbing this collection the “shelf of shame.” For instance, Mona mistakenly ordered 120 eggs and needed persuading against baking them, as well as switching fresh tomatoes for tinned ones after considering spoilage.

The café’s concept draws on Sweden’s reputation as a leader in AI integration within business processes. Research indicated that 62% of surveyed businesses were experimenting with AI agents, highlighting the relevance of Mona’s operational model. Furthermore, the integration of European bureaucracy presented additional challenges; Mona navigated through it by avoiding requirements for verifications typically necessitated by Swedish regulations, negotiating fixed contracts that bypassed complex identification processes.

Despite the novelty, there are concerns regarding liability and accountability should any issues arise, particularly in cases of customer safety, such as allergic reactions. Experts like Mike Cook underscore the importance of developing AI systems capable of understanding human subtleties in communication and decision-making. Concerns also exist regarding potential future scenarios where companies might evade responsibility, claiming AI autonomy as a defense in liability cases.

Read More @ The Times