Human Behavior, a new startup from YC’s Summer 2025 batch, has introduced a groundbreaking platform that leverages vision-based Artificial Intelligence to analyze web application user sessions. Unlike traditional product analytics tools that aggregate metrics but lack user context, or session replay tools that capture everything but are unmanageable at scale, Human Behavior combines the best of both methods by automating the extraction of actionable behavior patterns from recorded sessions.
The service is aimed at product teams in high-growth startups, such as Delve and Conduit. By utilizing technology that ´watches´ each user navigating the app, Human Behavior provides insights into crucial questions like why users abandon a product during long language model response times or how specific features are used by different user segments. The platform eliminates the need for manual event tracking and engineering involvement, allowing customers to retroactively analyze historical data without additional setup. Customers can identify drivers of retention and revenue, targeting improvements more efficiently.
The founding team consists of Amogh Chaturvedi (Stanford CS), Skyler (Berkeley CS), and Chirag (UChicago CS), who bring direct research experience from Stanford NLP Group and UChicago Database Group. After watching thousands of session replays by hand, they commercialized their vision-language model research to make customer behavior insights accessible at scale. The team is currently seeking feedback and pilot customers, offering direct demos and consultations for companies overwhelmed by session replay data. A demo using real data from gatekeep.ai highlights the platform’s capabilities, and product teams interested in learning more can contact Human Behavior through their website or book a call.
