Amogh chaturvedi It runs in a small sleep, but plenty of belief at 6am. It is groggy, apologetic for rearrangement, and still unfolds from a recent terrorism that includes a family member and an electric scooter.
In a matter of minutes, however, Stanford’s 20 -year abandonment is moving away, walking with how he and his co -founders sold a start at 19, landed on Y Combinator and brought $ 5 million for their next company, Human behavior.
Just a few months ago, human behavior bets that the AI vision can do whatever tools of analysis such as mixpanel and posthog have struggled with: to give companies a real understanding of how people use their products, including why they convert or hit.
Instead of relying on events with manual labels or clickstream data, human behavior claims that AI is monitoring the actual repetitions of user sessions and creating information, answering the most pressing questions of product groups without hours of instrument code.
The four -month YC launch closed the $ 5 million seed round in just two days (which becomes a rule for current YC companies), with supporters, including the general catalyst, Paul Graham, Vercel Ventures and Y Combinator.
“We could have done the game of financial engineering because we got more offers with higher valuations, but we didn’t want it,” the CEO said.
Chaturvedi met with his co -founders, Skyler Ji and Chirag kawediyaBoth 22, in a hacker house organized in 2023 as an excuse to live and build with friends after his first year at Stanford.
TechCrunch event
Francisco
|
27-29 October 2025
Their first start, the dough, was an e -commerce accounting tool. Like Chaturvedi, Ji left the college (leaving Berkeley) while Kawediya went to graduate.
Although the YC was initially skeptical of the dough market potential, the team was accepted in the spring of the accelerator accelerator this year with the assumption that they would eventually rotate, Chaturvedi says. They did it almost immediately after talking to every customer and asked about any other problems they faced.
The feedback was consistent: while Dough could show which products they sold or not, customers wanted to know why. Responding to this the required detailed data supported by behavior data, not just accounting reports.
In this new direction, the team sold dough for six numbers to employer.com, the same company that bought a bench, and went all-in for human behavior.
Kawediya explains that companies that use traditional details often need engineers to create event tracers for each button and clicks, combustion hours, sometimes weeks, time.
For a quick start, this is far from the ideal. “Even when you have this data, you are still stuck with the biggest question of how users really interact with your product so you can improve it,” he says.
The repetitions of sessions are not new, but until recently, computer models were not accurate enough to analyze them on a scale. Now it is, and human behavior does it to summarize and divide thousands of hours of material. “Why spend hours writing a code to keep track of clicking when we can watch the video?” Ji adds.
Today, customers of human behavior-mostly fast-moving series A and B newly formed-include daily brief emails that are highlighted which features were used, which have emerged and which users were stirred.
The founders call a meeting repeats a “untapped gold mine”. Right now, human behavior helps groups understand users and Squash errors. Over time, the same data set could supply automated QA and incorporate IT support. Their ambition is to make human behavior the datadog of the session repetition, turning dozens of products from the same core data.
Building with new ground technology is how the founders believe they will undertake more established players such as Mixpanel and Posthog. “For some of these companies, it may be difficult to reproduce what we have because their architecture cannot support the shift without starting,” Chaturvedi noted.
