What does Andon Labs do?
Andon Labs builds evaluations and benchmarks such as Vending-Bench, Butter-Bench, and Andon FM that test AI agents in simulated and real-world deployments, including running real vending machines, retail stores, and cafés. Founded in 2023 and backed by Y Combinator (W24), the San Francisco company works with labs including Anthropic, Google DeepMind, and OpenAI on AI control and safety research.