Startups that promise to predict how humans behave are having a golden moment. Over the past year, Simile has raised $200 million at a $2 billion valuation. Aaru raised $88 million at a $1 billion valuation. And AI startup Humans&, which announced a massive $480 million seed round at a $4.48 billion valuation in January, launched Persimmon to model human behavior.
The current state of human behavior prediction today relies heavily on large-scale language models (LLMs) that can be fine-tuned or encouraged to role-play as a target audience. But Miller Particle, a two-year-old San Francisco-based company, believes that approach is fundamentally broken.
“It’s like bringing super-soggy water to Niagara Falls,” says Abhibyakti Ahuja, co-founder and CEO of Mirror Particle, which provides brands with an AI engine that predicts consumer behavior and the reasons behind it. “LLM has been trained on hundreds of billions of data points. How much can you influence the behavior of LLM (by fine-tuning) with such a small amount of data? It’s still stuck in the past.”
Ahuja does not believe that LLMs see the world the same way humans do. “LLM models written language, but humans are made of visual perception, spatial reasoning, and social intelligence.” Relying on them, she says, means gaining insights based on things humans don’t notice, which is beside the point when trying to predict human behavior.
Mirror Particle takes a different approach. It’s building a foundational model, or as Ahuja describes it, a world model built from the ground up that simulates why humans behave the way they do and how human behavior changes over time.
“We don’t want to photograph static people,” Ahuja says. “We want to capture people as they change, and that means getting longitudinal data on how people are changing, what triggers them and to what extent.” If they aren’t changing, “that’s a signal, too,” she added.
Mirror Particle has already raised an angel round and says it is nearing the end of its first venture round. Next week, the company will also compete in TechCrunch’s prestigious startup competition, Startup Battlefield 200, at TechCrunch Disrupt 2026, October 13-15 in San Francisco.
The startup relies on a unique combination of data, including customer customer data, current events, pop culture, social media, and more, to model population segments and think of them as systems that evolve over time, tracking how motivation changes through experience. Much of the focus is on “overt behavior” – the actions people actually take rather than self-reported survey responses.
Like its competitors, Mirror Particle’s early go-to-market strategy focuses on where the budget for this type of insight already exists: market research and brand and product strategy. For example, Mirror could help beauty brands not only create better ad copy for cosmetics that appeal to Gen Z, but also determine whether that demographic wants the product.
“What if[the target audience]doesn’t want an eyeshadow palette?” Ahuja said. “If you want to sell a product to this market, teak is probably a better option.”
Mirror Particle’s predictive engine also helps brands make smarter decisions by providing customers with the “why” behind their current or future actions – motivations, constraints, and additional context to justify recommendations.
In an early pilot, a well-known pet food brand wanted to know what images to include on its packaging to increase sales. chicken? beef? vegetables? Mirror’s technology found that brands were asking the wrong questions. The image didn’t matter. The problem was that the brand was so well-recognized that it was considered mass-market and inexpensive, and sales would plateau until the perception issues were addressed.
“The way our model evolves is similar to how babies learn about the world,” Ahuja said, noting that babies move from vision to language to physical awareness to social intelligence.
That fundamental interest in modeling the human brain stems from Ahuja’s background studying neuroscience and computer science. Originally from India, she eventually studied at the University of Toronto, where she was inspired by AI pioneer Jeffrey Hinton’s contributions to neural networks.
After school, Ahuja ends up building a robot that builds other robots at Amazon Robotics. There she met co-founders Will Song and Thomson Yen. Song spent much of his career building sales personalization engines, while Yen focused on using deep learning to help AI agents learn how to understand human behavior.
The startup’s long-term vision is to become a “general layer for predicting human behavior,” moving from broader population-level analysis to individual-level insights.
“We need better human models to work with AI and with each other,” Ahuja said.
Check out Mirror Particle and many other innovative startups reviewed by the TechCrunch editorial team next week at Disrupt in downtown San Francisco. The winners of this year’s Startup Battlefield will be decided by VC judges on Thursday afternoon, October 15th.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
