Descriptions:
Stanford researcher Joon Park, creator of the influential Smallville generative-agent simulation paper, has taken his work far beyond a 25-character village. His new company, Simile, emerged from stealth in February 2026 with the ambition to model entire societies—drawing on demographic data, transaction logs, and scientific records to run simulations at population scale. The platform lets organizations ask concrete questions: How will a community respond to a tax change? How does news travel through a social network? What questions will analysts ask on an earnings call?
The company has serious backing. A $100 million seed round includes angel investors Andrej Karpathy (OpenAI co-founder), Fei-Fei Li (Stanford HAI co-director), Adam D’Angelo (OpenAI board member and Quora CEO), Guillermo Rauch (Vercel CEO), and Scott Belsky (Adobe Chief Strategy Officer). Enterprise clients CVS Health and Telra are already using the platform, and Simile reportedly predicted 8 out of 10 analyst questions during simulated earnings calls—giving companies a rehearsal tool for live investor scrutiny.
Wes Roth frames this as a potential paradigm shift from big data to big simulation: instead of mining stored historical records, organizations could run virtual experiments to forecast behavior. If accuracy continues to improve, large-scale surveys and retrospective data analysis may give way to on-demand societal simulations, with significant implications for market research, public health modeling, and social science.
📺 Source: Wes Roth · Published February 15, 2026
🏷️ Format: News Analysis
