AI digital twin startup Simile raises $100M in funding – News

Simile Inc., a startup that’s using artificial intelligence to generate digital twins of individuals, has raised $100 million in funding.

Lead investor Index Ventures announced the deal in a blog post published today. The venture capital firm was joined by several other institutional backers, including Bain Capital Ventures. AI pioneer Fei-Fei Li and OpenAI Group PBC co-founder Andrej Karpathy chipped in as well.

Before companies launch a new product, they often poll potential customers to collect feedback. They then use the feedback to refine the product before broadly rolling it out. Such research can be expensive and time-consuming to carry out, particularly when the target audience is difficult to reach. A new software startup, for example, may find it challenging to collect product feedback from Fortune 500 executives.

Simile is working to address the challenge. The company has built an AI model that uses data about individuals to simulate how they might respond to a new product, feature change and other business developments. Its initial customer roster includes CVS Health Corp. and Telstra Group, Australia’s largest mobile internet provider.

One of the tasks that Simile’s model promises to ease is testing user interface updates. Developers can evaluate how simulated  users respond to a change before rolling out it to customers.

According to Simile, the model can also help executives at publicly traded companies prepare for earnings calls. Chief Executive Joon Sung Park told Bloomberg that the model correctly forecast eight out of 10 analyst questions on a simulated call.

Park co-founded Simile with fellow computer scientists Michael Bernstein and Percy Liang. The trio previously developed Smallville, a simulated environment that demonstrated AI agents can be used to study human behavior. Smallville placed 25 agents in a video game setting and determined that they can simulate the behavior of not only individuals but also groups.

Smile reportedly trained its model on information collected from interviews with hundreds of people. Additionally, the training dataset included transaction logs and text from scientific journals. The model took seven months to develop. 

“Simile partners with real people to build high-fidelity models of how each of them lives and makes decisions,” Index Ventures partner Shardul Shah wrote in the blog post announcing Simile’s raise. “These digital twins are orchestrated to answer what real people will do and why. It turns out that there is a staggering range of applications and organizations that want to ask those questions.”

Buyer behavior is not the only task that startups are looking to streamline with AI-generated simulations. In 2024, Simile investor Li launched a company called World Labs Inc. to build models that generate three-dimensional virtual environments. Those environments lend themselves to tasks such as training industrial robots. 

Image: Simile