Home›Companies›Artificial Societies

Artificial Societies

Y Combinator LogoW25
Active
artificial-intelligence
machine-learning
saas
b2b
market-research
San Francisco
 
https://societies.io

We use AI to simulate entire human societies

Artificial Societies uses AI to simulate large groups of people and how they interact. Companies like Anthropic and 11x are using Artificial Societies to predict how their marketing and content will perform in a simulation of their target customers, before launching in the real world. This is a cracked team: James left rural China at age 14, got into Cambridge, turned down a CS PhD to do startups, and - in his spare time - lead the largest ever study of an LLM society. James is joined by Patrick, an applied behavioural scientist who has run over 200 real-world experiments for businesses including Fortune 500 companies. We joined forces through a shared frustration that no one really understands large scale human behaviours: traditional market research is not only slow and expensive, but fundamentally flawed - it misses how people influence each other. Humans are not isolated individuals, we are social animals. We began by simulating a network of 1,000 investors, and used it to raise a pre-seed and get into YC. Since releasing this publicly as “Wave”, startup founders have run over 3000 simulations. We have since released "Reach", where anyone can simulate their own LinkedIn audience and test how their posts will perform before posting for real. Artificial Societies is doing for simulations what ChatGPT did for LLMs - making them accessible to everyone. As AI advances, we will create representative models of entire human societies. Our vision is a world where all content, products, and policies are first simulated in an Artificial Society. We are on a mission to create Artificial Collective Intelligence - humans innovate in collectives, we want to scale collective innovation with ACI.
Artificial Societies
Founded:2024
Team Size:3
Status:
Active
Location:San Francisco
Group Partner:Jared Friedman
   
Active Founders

James He, Founder

CEO at Artificial Societies. Ex-Cambridge computational social science. Ex Yonder. At 16, James was offered $600k to fund his ed-tech startup but chose to attend Cambridge instead. After graduating, James turned down a PhD offer to join Yonder - a credit card startup - and built the entirety of their lending ML systems. He then started his own scholarship at age 21, and led the largest-scale study of an LLM society in his spare time. He co-founded Artificial Societies in Oct 2024.
James He
James He
Artificial Societies
 

Patrick Sharpe, Founder

Leading Product at Artificial Societies. Background in applied behavioural science. Patrick completed an MSc in behavioural econ, and left academia to lead behavioural science and CX projects at iptiQ - an insurtech startup owned by Swiss Re, the largest Reinsurer in the world. He joined Unconventional Wisdom, a specialised behavioural science consultancy, and scaled the client-base 3x within a year, before co-founding Artificial Societies in 2024.
Patrick Sharpe
Patrick Sharpe
Artificial Societies
 
Company Launches
Artificial Societies - Simulating entire human societies with AI
See original launch post ›

Hello World, we’re James and Patrick - cofounders of Artificial Societies 👋

We use AI to simulate large groups of people and how they interact. Our first product, Reach, is a simulation of your actual LinkedIn audience, where you can see if your posts will go viral before you launch them. It’s available now at societies.io

Our Story

We’re both behavioral scientists, but took different paths - James used machines to predict humans, while Patrick ran experiments in large organizations.

One day in a London coffee shop, James used a simulation to recreate an experiment Patrick had spent two months on… in just two minutes. That same week, Patrick quit his consulting job to co-found Artificial Societies with James.

We started by simulating a network of 1,000 investors, and used it to raise a pre-seed and get into YC.

Since releasing this publicly as “Wave”, startup founders have run over 3000 investor simulations.

And after getting a taste for simulations, our batch-mates kept asking us the same thing over and over: “can you simulate my LinkedIn?”

So we made something people want.

Introducing Reach

For the first time in history, you can create an AI simulation of your own LinkedIn audience. Reach is your LinkedIn brand-building co-pilot, helping you test and refine your posts before you publish.

Reach in Action

Already, we can predict how many people will engage with a post at a 83% accuracy, compared with around 17% when asking ChatGPT.

This is the beauty of simulating not just individuals, but also how people connect and influence each other: humans are not isolated individuals, we are social animals.

What’s next?

Imagine creating thousands of copies of your own artificial society, and seeing how thousands of different content, product ideas, and even Launch YC posts would land in these parallel universes.

We plan to create representative models of entire human societies. Our vision is a world where all content, products, and policies are first simulated in an Artificial Society.

Our Ask 🙏

Reach is available now. Please sign up and share your feedback! Insights from Pro users will be immensely valuable to us ;)

If you’re interested in Reach for enterprise, or you want to use Reach for clients, send us a message at founders@societies.io