At 6 a.m., Amogh Chaturvedi is operating on minimal sleep but unwavering determination. He’s a touch groggy, apologetic for a last-minute reschedule, and still processing a recent scare involving a family member and an electric scooter.

Yet within minutes, the 20-year-old Stanford dropout sharpens his focus, walking through the remarkable journey that led him and his co-founders to sell their first startup at 19, join Y Combinator (YC), and raise $5 million for their latest venture: Human Behavior.

Launched only a few months ago, Human Behavior is placing a bold bet: vision AI can solve a problem that legacy analytics tools like Mixpanel and PostHog have long struggled with — giving businesses a true, actionable understanding of how users engage with their products, including the reasons behind conversions and customer churn.

Unlike traditional tools that rely on manually labeled events or clickstream data, Human Behavior says its AI analyzes real user session replays to generate insights. This lets product teams get answers to their most critical questions without spending hours writing and implementing tracking code.

The four-month-old YC startup closed its $5 million seed round in just two days — a pace that’s growing increasingly common among current YC-backed companies — with support from heavyweight investors including General Catalyst, Paul Graham, Vercel Ventures, and Y Combinator itself.

“We had the option to play the financial engineering game; we received multiple offers with higher valuations,” Chaturvedi, the company’s CEO, said. “But that wasn’t what we wanted — we’re focused on building something meaningful, not just chasing a higher number.”

Chaturvedi crossed paths with his co-founders — Skyler Ji and Chirag Kawediya, both 22 — at a hacker house he organized in 2023. The gathering started as a way to live and build alongside friends after his freshman year at Stanford, but it quickly evolved into the foundation of their professional partnership.

Their first startup, Dough, was a bootstrapped e-commerce accounting tool. Like Chaturvedi, Ji dropped out of college (leaving the University of California, Berkeley), while Kawediya stayed to complete his degree.

YC was initially doubtful about Dough’s market potential, Chaturvedi notes, but the team was accepted into the accelerator’s spring 2025 batch under the expectation that they would eventually pivot. That pivot came almost right away: After talking to every Dough customer and asking about other challenges they faced, the team realized a critical gap.

The feedback was uniform: While Dough could show customers which products were selling (and which weren’t), it couldn’t answer the bigger question: why. To address that, they needed analytics rooted in behavioral data — not just surface-level accounting reports.

With this new direction in mind, the team sold Dough for a six-figure sum to Employer.com (the same company that acquired Bench) and turned their full attention to building Human Behavior.

Kawediya, Human Behavior’s COO, explains the pain point with traditional analytics: Companies often need engineers to set up custom event trackers for every button click and user action, which eats up hours — sometimes even weeks — of valuable engineering time.

For fast-paced startups, this inefficiency is a major roadblock. “Even once you have that tracked data, you’re still left with the hardest part: figuring out how users actually interact with your product so you can improve it,” he says.

Session replays themselves aren’t new, but until recently, computer vision models lacked the accuracy to analyze them at scale. Today, that’s changed — and Human Behavior is leveraging this advancement to summarize and segment thousands of hours of user session footage. “Why waste hours writing code to track clicks when we can just let AI ‘watch’ the session videos?” Ji, the company’s CTO, adds.

Currently, Human Behavior’s clients — mostly fast-growing Series A and B startups — receive daily summary emails that highlight key insights: which product features are being used, which bugs have popped up, and which users have churned.

The founders refer to session replays as an “untapped goldmine.” Right now, Human Behavior helps teams understand user behavior and fix bugs, but the vision is broader: Over time, the same dataset could power automated quality assurance (QA) and embedded IT support. Their long-term goal? To make Human Behavior the “Datadog of session replay” — a central platform that spins off dozens of specialized products from its core behavioral data.

The team believes building their technology from the ground up with vision AI is the key to competing with established players like Mixpanel and PostHog. “For some of these older companies, replicating what we’re doing would be really tough,” Chaturvedi noted. “Their existing architectures aren’t built to support this shift — they’d basically have to start over.”