Simile builds a foundation model for human behavior — what it calls "the simulation layer" for enterprise decisions. Companies feed it structured interview data, behavioral signals, and past choices; Simile creates high-fidelity AI agents (one per real person) that can be assembled into populations and queried at scale. The result: a business can test a product launch, earnings call script, pricing change, or policy before committing real dollars — getting not just a prediction but a distribution of likely responses with a confidence score attached to each. Current enterprise use cases include rehearsing investor Q&A, modeling patient reactions to benefits changes, and stress-testing marketing campaigns. Customers include CVS Health, Gallup, Wealthfront, Deloitte, Suntory, Telstra, and Banco Itaú. The founding team — Joon Sung Park, Percy Liang, and Michael Bernstein — authored the seminal Stanford "Generative Agents" and "Social Simulacra" papers that created this research field.