Sujay Choubey spent nearly six years as an early-stage investor in India, studying what separated the companies that scaled from the ones that stalled. Then he did something investors rarely do. He left the fund to build inside one of the categories he had spent years analyzing. His path traces how a generation of operators is moving toward the front lines of the AI agent economy.
When Sujay Choubey graduated from the Indian Institute of Technology, Delhi in 2016, his early path pointed nowhere near venture capital or artificial intelligence. As an undergraduate in electrical power engineering, he co-authored research applying machine learning to the forecasting of wind power output. It was an early brush with the statistical methods that now sit beneath much of the technology he works with today. His first full-time role was not at a startup but at the Quality Council of India, an autonomous body set up by the Indian government and industry bodies, where he spent close to three years managing national scale efficiency projects for ministries spanning commerce, railways, and oil and gas.
That early training, learning to measure what works and to steer complex systems toward a result, would shape the next decade of his career, though rarely from the same seat at the table.
Learning to Spot What Wins
In 2019, Mr. Choubey moved into venture capital, joining Blume Ventures, one of India’s most active early-stage firms, as an investment analyst. There he worked around companies building in hard, capital-intensive categories, including the electric vehicle maker Euler Motors, the battery-swapping network Battery Smart, and Bureau, a risk and identity infrastructure company. The work was less about chasing trends than about understanding what made a young company durable.
In 2021 he joined the investment team at Sequoia Capital’s India practice, the firm that in 2023 spun off from its American parent and rebranded as Peak XV Partners, one of the largest venture firms across India and Southeast Asia. Over nearly four years with the firm, Mr. Choubey concentrated on enterprise software, developer tools, and applied artificial intelligence, working around companies such as Enterpret, a customer feedback analytics platform, RapidCanvas, a no-code enterprise AI platform, and Requestly, a developer tool for API mocking. The pattern in those bets would prove telling.
“Sitting on the investing side, you start to see patterns before they are obvious. You watch the same problem show up across twenty companies and you realize it is not a company problem, it is an infrastructure problem. The more time I spent around applied AI, the more convinced I became that the biggest gaps were not in the models. They were in the plumbing that lets those models actually do useful work.”
Leaving the Fund to Build
In May 2025, Mr. Choubey made the less conventional move. Rather than continue evaluating companies from the outside, he joined one as an operator, taking on a growth role at Composio, a San Francisco and Bengaluru based startup building integration infrastructure for AI agents. Two months later, Composio announced a $25 million Series A round led by Lightspeed Venture Partners, bringing its total funding to about $29 million.
The company sits at one of the busiest intersections in artificial intelligence. Its platform lets AI agents connect to the everyday software that businesses already run on, from Gmail and GitHub to Salesforce and Slack, and it handles the authentication and tool calling that developers would otherwise build by hand. By the time of its Series A, Composio reported more than 100,000 developers on the platform and over 200 paying companies, among them the enterprise search company Glean.
“Analyzing a company and building one are completely different muscles. As an investor you are paid to have an opinion. As an operator you are paid to be right, and then to go and make it happen with a real team and a real deadline. I wanted to be closer to the building.”
Inside the Agent Economy
At Composio, Mr. Choubey leads growth, a mandate he describes as spanning go-to-market strategy, developer adoption, partnerships, and hiring. He has become a visible figure in San Francisco’s fast-growing community of agent builders, presenting the company’s connector tooling at developer gatherings across the city. Much of his focus has centered on work around the Model Context Protocol, an emerging standard for how AI agents discover and use external tools.
Composio has positioned that work around a universal gateway that gives agents a single, reliable way to reach thousands of applications, a problem that grows more pressing as companies move from AI demonstrations to systems that take real actions on their behalf. It is a market moving faster than anything Mr. Choubey encountered as an investor.
“One enterprise software company told us their own customers had connected their accounts through us and basically stopped logging in. The work was still getting done, just through automation rather than a dashboard. I think every software company is going to have a version of that moment in the next couple of years. That is the shift we are building for.”
A Pattern Across Roles
What stands out about Mr. Choubey’s path is not any single role but the thread connecting them. Whether he was managing efficiency projects for government ministries, evaluating early-stage companies at Peak XV, or now helping a young infrastructure company find its market, the work has circled the same question: what does it actually take to make a complex system work reliably at scale.
Now, at Composio, he is applying that accumulated vantage point to a category that is still being defined. The market for AI agent infrastructure does not reward operators who learn slowly. For Mr. Choubey, years spent studying what makes companies succeed, from inside government, venture capital, and now a startup, mean he arrives at the problem already knowing where to look.