Vision AI Startup Raises $5M to Study Online Behavior

A Young Team's Vision for Understanding User Behavior

Amogh Chaturvedi, despite operating on minimal sleep, demonstrates strong resolve at 6 a.m. He expresses regret for a schedule adjustment and recounts a recent incident involving a family member and an electric scooter.

However, the 20-year-old, formerly enrolled at Stanford, quickly focuses, detailing how he and his co-founders successfully sold a previous startup at age 19, gained acceptance into Y Combinator, and secured $5 million in funding for their current venture, Human Behavior.

Introducing Human Behavior and its Core Technology

Launched recently, Human Behavior is pioneering the use of vision AI to address limitations found in traditional analytics platforms like Mixpanel and PostHog. The company aims to provide businesses with a genuine understanding of user product interaction, including the factors driving conversion and churn.

Rather than relying on manual event tagging or clickstream data, Human Behavior leverages AI to analyze actual user session recordings. This process generates actionable insights, answering critical product team questions without extensive code instrumentation.

The startup, only four months old, rapidly closed a $5 million seed funding round in a mere two days – a trend increasingly common among current Y Combinator participants. Investors include General Catalyst, Paul Graham, Vercel Ventures, and Y Combinator itself.

“We bypassed pursuing complex financial strategies, despite receiving offers with potentially higher valuations,” the CEO stated. “Our focus remained on building a sustainable and impactful company.”

The Team's Origins and Previous Venture

Chaturvedi connected with his co-founders, Skyler Ji and Chirag Kawediya, both 22 years old, at a collaborative living and building space he organized in 2023 following his first year at Stanford.

Their initial company, Dough, was an e-commerce accounting solution developed through self-funding. Ji, similar to Chaturvedi, chose to leave his university (Berkeley), while Kawediya completed his degree.

Although initially hesitant about Dough’s market viability, Y Combinator admitted the team to its accelerator program with the expectation of a strategic pivot, according to Chaturvedi. This pivot occurred swiftly after extensive customer interviews revealed unmet needs.

Customers consistently expressed a desire to understand the *reasons* behind sales trends, not just the trends themselves. Addressing this required behavioral analytics, extending beyond simple accounting reports.

From Dough to Human Behavior

Driven by this insight, the team sold Dough to Employer.com – the same entity that acquired Bench – and dedicated themselves fully to Human Behavior.

Kawediya highlights the inefficiencies of traditional analytics, which often necessitate engineers to implement event trackers for every user interaction, consuming valuable development time, potentially spanning weeks.

For rapidly growing startups, this represents a significant drawback. “Even with the data collected, the fundamental question of how users genuinely interact with the product to facilitate improvement remains,” he explains.

The Power of Session Replays and Vision AI

Session replays are not a novel concept, but the recent advancements in computer vision have enabled their analysis at scale. Human Behavior is leveraging this capability to summarize and categorize extensive footage.

“Why dedicate hours to coding click tracking when we can simply observe the user experience?” Ji proposes.

Currently, Human Behavior provides its customers – primarily Series A and B startups – with daily email summaries detailing feature usage, identified bugs, and user churn patterns.

Future Vision and Potential

The founders describe session replays as a largely untapped resource. Currently, Human Behavior aids teams in understanding users and resolving bugs. However, they envision the same data powering automated quality assurance and integrated IT support in the future.

Their ultimate goal is to establish Human Behavior as the leading platform for session replay data, fostering the development of numerous products from a single core dataset.

The founders believe that building upon cutting-edge technology from the outset will allow them to compete effectively with established players like Mixpanel and PostHog. “Some existing companies may find it challenging to replicate our capabilities due to architectural limitations that would necessitate a complete overhaul,” Chaturvedi concludes.