A field report from the Build Small Hackathon on a tiny multi-agent economy simulation powered by a 3-billion-parameter model. The project demonstrates that small models can enable real-time multi-agent simulations when designed with engineered scarcity and careful prompting, revealing both the reliability and limitations of small models.

AI agents need secure execution environments. LangSmith Sandboxes provide hardware-virtualized microVMs, giving each agent a full computer with fast startup and persistent state, enabling code generation, data analysis, CI workflows, and more.