Simile builds AI digital twins of people, generative agents that simulate human decisions and predict how customers react to products and pricing. The launch pairs a foundation model of human behavior, trained on datasets of what real people say and do, with a predictive accuracy score that estimates how reliable a given simulation is before customers act on it. It is aimed at enterprises making high-stakes calls on product, pricing, marketing, and policy, where the current alternatives are generic personas, one-off surveys, and guesswork.
The company grew out of Stanford research on generative agents. Simile is led by co-founders Joon Park, Michael Bernstein, Percy Liang, and Lainie Yallen, all of whom have academic ties to Stanford University. It was founded by Stanford PhD graduate Joon Sung Park, whose dissertation involved a project called "Smallville" in which AI agents carried on simulated human lives , and Percy Liang helped popularize the term foundation model through his work at Stanford's CRFM. Early enterprise use cases span improving product and marketing concepts, rehearsing earnings calls, modeling litigation outcomes, and testing policy changes , with named customers including CVS Health, Gallup, Wealthfront, and Suntory Beverage & Food.
This launch marks Simile's Series B of $200M at a $2B valuation, led by Greenoaks with participation from Index Ventures, Hanabi, A*, Bain Capital Ventures, CVS Health Ventures, and Definition. It arrives roughly five months after the company's $100M Series A, and signals a push to scale the platform from bespoke enterprise engagements toward a broader ambition of simulating populations accurately enough for operators to experiment on the future before committing to a decision.
Comments (10)
Sign in to join the discussion.
K
Kwesi BoatengJul 30, 2026
Simulating 8 billion people accurately is a wild sentence to put in a Series B announcement. I respect the audacity of promising the entire population as a feature.
P
Priyanka R.Jul 30, 2026
$200M at $2B means Greenoaks is underwriting a lot of GPU bills. Curious what the revenue shape looks like when your product is a synthetic focus group.
A
Adaobi NwosuJul 30, 2026
CVS Health Ventures on the cap table is the most interesting name here. Someone at CVS is very tired of running actual pilots.
M
milosJul 30, 2026
The launch video cuts feel like a Nolan trailer for a company that runs marketing surveys. Whoever edited that earned their invoice.
T
Tomoe H.Jul 30, 2026
Every simulation startup I've seen ends up being three grad students in a trenchcoat calling GPT-4. Would love to know what's actually novel under the hood.
L
Leo HartvigJul 30, 2026
Is there an API? Asking because 8 billion agents sounds fun until I hit a rate limit trying to A/B test a checkout button.
Y
Yuka SerizawaJul 30, 2026
Population-scale agent modeling has a whole literature going back to Epstein and Axtell. Would genuinely like to know how they benchmark accuracy against real behavioral data.
S
Sanaa IdrissiJul 30, 2026
If you're staffing research eng post-B, I know a computational social scientist who has been waiting for exactly this job to exist.
R
Raffi OhanianJul 30, 2026
Cool, but who actually buys this on Monday morning? The go-to-market for 'simulate humanity' is either every Fortune 500 or literally no one.
D
Deniz K.Jul 30, 2026
Please tell me the simulated version of me is landing more dates than the real one. I'll pay for that seat.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Simile
Simulate every decision at scale
Simulate every decision at scale
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Simile builds AI digital twins of people, generative agents that simulate human decisions and predict how customers react to products and pricing. The launch pairs a foundation model of human behavior, trained on datasets of what real people say and do, with a predictive accuracy score that estimates how reliable a given simulation is before customers act on it. It is aimed at enterprises making high-stakes calls on product, pricing, marketing, and policy, where the current alternatives are generic personas, one-off surveys, and guesswork. The company grew out of Stanford research on generative agents. Simile is led by co-founders Joon Park, Michael Bernstein, Percy Liang, and Lainie Yallen, all of whom have academic ties to Stanford University. It was founded by Stanford PhD graduate Joon Sung Park, whose dissertation involved a project called "Smallville" in which AI agents carried on simulated human lives , and Percy Liang helped popularize the term foundation model through his work at Stanford's CRFM. Early enterprise use cases span improving product and marketing concepts, rehearsing earnings calls, modeling litigation outcomes, and testing policy changes , with named customers including CVS Health, Gallup, Wealthfront, and Suntory Beverage & Food. This launch marks Simile's Series B of $200M at a $2B valuation, led by Greenoaks with participation from Index Ventures, Hanabi, A*, Bain Capital Ventures, CVS Health Ventures, and Definition. It arrives roughly five months after the company's $100M Series A, and signals a push to scale the platform from bespoke enterprise engagements toward a broader ambition of simulating populations accurately enough for operators to experiment on the future before committing to a decision.
Comments (10)
Simulating 8 billion people accurately is a wild sentence to put in a Series B announcement. I respect the audacity of promising the entire population as a feature.
$200M at $2B means Greenoaks is underwriting a lot of GPU bills. Curious what the revenue shape looks like when your product is a synthetic focus group.
CVS Health Ventures on the cap table is the most interesting name here. Someone at CVS is very tired of running actual pilots.
The launch video cuts feel like a Nolan trailer for a company that runs marketing surveys. Whoever edited that earned their invoice.
Every simulation startup I've seen ends up being three grad students in a trenchcoat calling GPT-4. Would love to know what's actually novel under the hood.
Is there an API? Asking because 8 billion agents sounds fun until I hit a rate limit trying to A/B test a checkout button.
Population-scale agent modeling has a whole literature going back to Epstein and Axtell. Would genuinely like to know how they benchmark accuracy against real behavioral data.
If you're staffing research eng post-B, I know a computational social scientist who has been waiting for exactly this job to exist.
Cool, but who actually buys this on Monday morning? The go-to-market for 'simulate humanity' is either every Fortune 500 or literally no one.
Please tell me the simulated version of me is landing more dates than the real one. I'll pay for that seat.