Mirror Particle is building a world model of human behavior
San Francisco startup Mirror Particle is building a foundation model from scratch to simulate human behavior, competing at TechCrunch Disrupt 2026.
Stock photo for illustration only, not from the actual event
- Mirror Particle builds a foundation model to simulate human behavior from scratch
- Founded by former Amazon Robotics team members Ahuja, Will Song, and Thomson Yen
- Competing in Startup Battlefield 200 at TechCrunch Disrupt 2026
- Focuses on revealed behavior rather than traditional survey responses
The current status quo for predicting human behavior relies heavily on large language models prompted to role-play as target demographics. However, two-year-old, San Francisco-based Mirror Particle believes this approach is fundamentally flawed because LLMs are designed to model written text rather than human perception.
Ahuja argues that humans possess visual perception, spatial reasoning, and social intelligence, elements that LLMs lack. Consequently, Mirror Particle is taking a different approach by building a foundational world model from scratch to simulate why humans act the way they do and how their behaviors evolve over time.
Stock photo for illustration only, not from the actual event
The startup has already secured an angel round and is close to closing its first venture round. Furthermore, the company is competing in the Startup Battlefield 200 startup competition at TechCrunch Disrupt 2026, held in San Francisco from October 13 to 15.
Developing a world model for human behavior represents a significant departure from standard language models. By attempting to simulate psychological, social, and environmental contexts rather than just textual patterns, such systems aim to provide deeper insights into consumer decision-making while mitigating survey bias.
Mirror Particle utilizes a proprietary combination of client data, current events, pop culture, and social media to model demographic segments as evolving systems. The core focus remains on revealed actions rather than self-reported survey answers.
"What if the target demographic doesn't want eyeshadow palettes? Maybe blush is a better option to go for if you want to sell a product to this market."
Ahuja
The startup's initial market strategy targets brand and product strategy. For instance, in an early pilot involving a pet food brand, Mirror's technology determined that packaging imagery was not the issue; rather, the brand suffered from a mass-market perception hurdle that hindered sales growth.
Looking ahead, Mirror Particle envisions becoming the general layer for anticipating human behavior, scaling insights from broad population analyses down to individual levels.
Source: TechCrunch
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment