Simile

Simile is an artificial intelligence startup that develops AI models to simulate and predict human behavior. The company is building a foundation model trained to forecast how individuals and populations will respond to various scenarios, such as new products, policy changes, or corporate announcements. Simile’s platform creates simulations using "generative agents," which are AI models designed to replicate the decision-making patterns, preferences, and behaviors of real people. This technology allows organizations to test decisions and understand potential outcomes before implementing them in the real world.

The company was founded by a team with deep academic roots at Stanford University: CEO Joon Sung Park, Chief Product Officer Michael Bernstein, Chief Scientist Percy Liang, and Chief Customer Officer Elaina Yallen. Park's Ph.D. research at Stanford, which introduced the concept of "generative agents," serves as a cornerstone of the company. A notable 2023 research project called "Smallville" demonstrated the potential of this technology by creating a virtual town of 25 AI agents that exhibited emergent social behaviors without explicit instructions. Liang is credited with coining the term "foundation model" and is the director of Stanford's Center for Research on Foundation Models (CRFM), while Bernstein is a co-author of the influential ImageNet project that was pivotal for computer vision.

Simile emerged from a seven-month stealth period in February 2026, announcing a $100 million Series A funding round. The round was led by Index Ventures, with participation from Bain Capital Ventures, A*, and Hanabi Capital, along with prominent AI researchers like Fei-Fei Li and Andrej Karpathy. The company's business model targets enterprise clients, offering its simulation platform to help with strategic decisions. Use cases include rehearsing for earnings calls, where the model has been used to predict analyst questions with high accuracy, modeling litigation outcomes, and guiding product strategy. Simile’s training data incorporates hundreds of in-depth interviews, historical transaction data, and findings from behavioral science journals to ground its AI agents in real-world human reasoning. Clients such as CVS Health, Telstra, Wealthfront, and Gallup are using the platform to test inventory strategies, understand customer segments, and create digital polling panels.

Keywords: human behavior simulation, generative agents, foundation model, predictive AI, decision intelligence, AI simulation, behavioral analytics, agent-based modeling, computational social science, AI forecasting, enterprise AI, market simulation, consumer behavior, strategic planning, risk management, product testing, AI research, Stanford spinout, Joon Sung Park, Percy Liang