$200 million to simulate human decisions

The Simile platform models synthetic consumer reactions—a method for testing campaigns and products prior to market research.

Five months after emerging from stealth, Simile has closed a Series B round of over $200 million at a $2 billion post-money valuation. Greenoaks led the round, with participation from Index Ventures, Hanabi, A, Bain Capital Ventures, Factory, CVS Health Ventures, and new investor Definition. This funding follows a $100 million Series A, bringing the company's publicly disclosed funding to over $300 million in just a few months.

Concretely, Simile is a simulation platform designed for enterprises. It creates synthetic populations composed of agents representing various consumer, patient, or user profiles. An organization can subject a product, campaign, customer journey, offer, or strategic decision to them, and then observe their likely reactions before conducting a real-world study or experiment. The system specifically aims to highlight preferences, barriers, differences between groups, and the potential effects of a change.

Simile claims to have grown its revenue fivefold since its launch, run tens of millions of simulations for large enterprises, and developed a second model tasked with estimating the confidence level of each result. CVS Health, Wealthfront, Deloitte, and Gallup are among the clients cited. The team now exceeds 50 people spread across several countries.

The project builds on several research works conducted at Stanford by Joon Sung Park, Percy Liang, and Michael Bernstein. A study published in 2024 created agents based on two-hour interviews conducted with 1,052 individuals. On the General Social Survey, their responses matched those of the participants with a fidelity equivalent to 85% of the consistency observed when those same individuals answered a second time two weeks later. This result therefore measures proximity to the stability of human responses, rather than a general ability to predict 85% of behaviors.

At CVS Health, the simulations rely on 2.9 million consented responses from over 400,000 participants, covering more than 200 behavioral scenarios. They have notably been used to test care pathways, patient messaging, service engagement levers, and various positioning hypotheses before moving forward with real-world studies or experiments.

The new funding is intended to support model training, accelerate the learning loop, and expand simulations to larger groups in which consumers, companies, competitors, and institutions can interact. Simile is thus no longer just seeking to replicate an individual response, but to anticipate the successive effects of a decision within a complex human system.