Human behavior prediction is a crowded, well-funded space right now: Simile raised $200 million at a $2 billion valuation, Aaru raised $88 million at $1 billion, and Humans& launched Persimmon after announcing a $480 million seed round at a $4.48 billion valuation. The status quo for human behavior prediction today relies heavily on large language models that are prompted or fine-tuned to role-play a target demographic. Mirror Particle, a two-year-old San Francisco startup, argues that this status quo is fundamentally broken. CEO Abhivyakti Ahuja compares it to “bringing a super soaker to Niagara Falls”: LLMs are trained on hundreds of billions of data points, so fine-tuning with a small amount of data can’t meaningfully shift their behavior. She also argues that LLMs model written language, not the visual perception, spatial reasoning, and social intelligence that shape human decisions.

Mirror Particle is instead building a foundation model — a “world model” built from scratch that simulates why humans do what they do and how that behavior changes over time. Ahuja says the goal is not to capture a static person but a changing one: the company captures longitudinal data on how people are shifting, what triggers are changing them, and to what degree; a lack of change is also a signal. The model combines clients’ customer data with current events, pop culture, social media, and other inputs to represent a demographic segment as an evolving system. The company emphasizes “revealed behavior” — what people actually do — rather than self-reported survey answers.

Mirror Particle‘s initial focus is where budgets already exist: market research and brand and product strategy. The engine can help a beauty brand decide whether a demographic wants a product at all, not just how to write ad copy for it. In an early pilot for a well-known pet food brand, Mirror found that packaging imagery (chicken, beef, vegetables) was the wrong question; the real constraint was brand perception, since the brand was so recognizable it was considered mass market and cheap, and sales would plateau until that changed. The prediction engine also returns the “why” behind current or future behavior — motivations, constraints, and context.

The startup has raised an angel round and is close to closing its first venture round. It is competing next week in Startup Battlefield 200 at TechCrunch Disrupt 2026 in San Francisco on October 13-15. Ahuja studied neuroscience and computer science at the University of Toronto, was inspired by Geoffrey Hinton’s contributions to neural networks, then worked at Amazon Robotics, where she met co-founders Will Song and Thomson Yen. Mirror Particle‘s long-term vision is to be the “general layer for anticipating human behavior,” moving from population-level analyses to individual-level insights. As Ahuja puts it, “We just need a better model of humans if we’re going to work alongside AI and with each other.”