Artificial Societies has raised $5.35M to replace the guesswork in market research
What if you could build a replica of real world societies that could test how people respond to decisions, products, and policies before they’re even made?
That’s the question co-founders James He and Patrick Sharpe asked themselves before co-founding Artificial Societies.
James had left rural China at fourteen, enrolling at Cambridge before turning down a computer-science PhD to build things. Patrick Sharpe spent years as an applied behavioral scientist, running more than 200 real-world experiments for large companies.
Together they created a crowd-modeling platform that generates AI clones with traits, preferences and biases that mirror real people.
Before pitching to investors, they simulated their pitch to 1,000 clones, allowing them to rehearse questions and tightened their pitch, helping them raise and secure a place in YC’s W25 batch where they shipped their first public experiments.
One of which was Wave, their earlier investor simulation, which allowed founders to run thousands of their own simulations. Soon after users began to ask , “Can you run this on my LinkedIn audience?” The team responded with Reach, simulating LinkedIn audiences and pre-tests posts before they go live.
By mid-2025, more than 15,000 users had run 100,000+ simulations on the platform, testing hooks, thumbnails, feature framing, and crisis lines. In August, the company raised $5.35 million led by Point72 Ventures, with earlier checks from angels tied to DeepMind, Strava, and Sequoia Scout.
For a decade, data mostly told us what already happened. Artificial Societies wants to specialise in what might happen next. If the company can continue proving tha t what wins in rehearsal more often wins onstage, they’ll move from novelty to necessity.
Allowing you to walk into launch having already heard the room.
