Indexing Humanity: A Fireside With the Simulation Company | Simile | RAISE Summit 2026

• Simile is a company creating a foundational model of human behavior to simulate individual subpopulations for enterprise clients, including CVS, Amex, Walmart, and Fortune 500 companies.
• The founding team consists of Jun (CEO), Michael Bernstein, Percy Leung, and Lane Yellen, all former researchers from Stanford's computer science department.
• The team co-founded the "foundational model" concept and previously led the 2023 "Generative Agents" research, which created "Smallville," a simulated town of 25 agents exhibiting daily routines and relationships.
• In late 2024, the team published a paper claiming their models predict human behavior 85% as accurately as the individuals themselves.
• Simile has established strategic partnerships with Gallup and Dynata, granting access to a panel of approximately 20 million human individuals globally for data training.
• The company differentiates its models from general-purpose AI labs (e.g., OpenAI, Anthropic) by prioritizing human imperfection, biases, and subjective taste over objective reasoning or coding efficiency.
• Simile views its technology as a tool for augmenting human decision-making rather than replacing labor, aiming to inject human values into automated decision systems.

Market Entry and Use Cases

• Early traction was driven by inbound interest from Fortune 500s after executives observed the technology could replicate complex studies (e.g., McKinsey reports) costing millions and taking months in a single demo session.
• The initial market wedge is market research, specifically for A/B testing, concept testing, and focus groups.
• Use cases have expanded horizontally to product development (simulating user interaction with prototypes) and corporate communications (simulating earnings calls for CEO preparation).
• Customers report being bottlenecked by human resource constraints and time; Simile addresses this by offering simulations that can run without the months-long delays of traditional studies.

Strategic Philosophy and Risk Management

• Simile operates on a four-stage simulation hierarchy:
  • Stage 1: Individual level static simulation (automating surveys and experiments).
  • Stage 2: Individual level dynamic simulation (temporal dimension, e.g., simulating 10-15 minutes of product usage).
  • Stage 3: Multi-agent simulation (agents interacting to generate immersion phenomena).
  • Stage 4: Society-wide simulation (simulating entire markets and downstream societal implications).
• The company builds trust through extreme transparency, providing end-to-end data loops to show users how inputs are leveraged into outputs, avoiding "black box" opacity.
• Simile adheres to two core ethical principles: informed consent for represented individuals and the autonomy for individuals to opt out of their data being used in simulations.
• The company is currently navigating the nascent stage of the industry, focusing on setting norms and preventing misuse similar to the early days of large language models.

Future Outlook and Contact

• Simile believes the cost of content creation will approach zero, creating a need for mechanisms to ensure automated outputs align with human values and taste.
• The company plans to focus on modeling specific target populations rather than verticals, noting that diverse sectors (retail, finance, CPG) often share the same consumer demographics.
• The company is based in California and was recently supported by Index Ventures in a Series A round.
• Interested parties can contact Simile via email at joo.n@simile.ai or visit simile.ai.