Andy Li

Andy Li is co-founder and CEO of Respan, an AI engineering platform for tracing, evaluating, and improving agents in production. With classmates from the University of Illinois Urbana-Champaign he started the company as Keywords AI, joined a YC Winter 2024 batch, and later rebuilt and rebranded around agent observability after customers said routing alone was not the hard problem.

From AirPods engineering to YC

Li studied mechanical engineering and economics at UIUC, then worked at Apple as a product design engineer on parts of AirPods 4. In a Pulse 2.0 interview he said he left school to build full-time with friends he already trusted: "I studied mechanical engineering and economics at UIUC, where I met my co-founders and some of my closest friends." The early company lived in Bay Area hacker houses in Alameda, which he still cites as the culture that made product decisions and late-night debugging feel like one continuous conversation.

Keywords AI becomes Respan

Keywords AI began as infrastructure for dynamic LLM routing across model providers. During YC, Li kept hearing that teams shipping multi-step agents could see that quality dropped or costs spiked, but could not explain why. "The bigger problem was not routing itself," he told Pulse 2.0. "AI teams were starting to ship complex agents into production, but they had almost no visibility into how those agents behaved, why they failed, or how to improve them over time." Respan now combines production tracing, evaluations, prompt workflows, and a gateway, and the company has said it raised a $5 million seed led by Gradient with participation from Y Combinator and others.

On closing the agent loop

Li's public framing is that dashboards are not enough once agents plan, call tools, and use memory across turns. He wants Respan to close a loop: show what happened, measure quality, catch regressions, and help teams improve the system. In the same interview he put the longer bet plainly: "Long term, our goal is to become the default observability and improvement layer for every team shipping AI agents in production."