Robert Kim

Robert Kim is co-founder and CEO of Laminar, an open-source observability platform for long-running AI agents. With cofounder Dinmukhamed Mailibay he grew up in Kazakhstan, later worked on infrastructure at Palantir and Bloomberg, and joined a YC Summer 2024 batch. In March 2026 Laminar announced a $3 million seed led by Atlantic with participation from Y Combinator and angels including OpenTelemetry co-creator Ben Sigelman.

Kazakhstan to London agent tooling

Kim and Mailibay studied together at KAIST and worked side by side in London before founding Laminar. Kim has said his Palantir internship included building a semantic search package used by internal AI teams, and at Bloomberg he worked on scaling a market tick processing pipeline. That systems background shows up in how Laminar talks about agents that run for half an hour and produce hundreds of decisions instead of a single request-response call.

Observability built for long agent runs

On the company blog and in coverage of the seed round, Kim frames the gap plainly. Existing tools were designed for short LLM calls. Agents that browse, call tools, and fail late in a run leave teams staring at thousands of spans. In an Atlantic write-up he put the product bet this way: "When your agent fails 40 minutes into a task, today's tools show you a wall of thousands of spans and say 'good luck.' We built Laminar so you can pinpoint the exact decision that went wrong and rerun from that point."

Debugger, Signals, and open source

Laminar's public launch materials emphasize a debugger that restores prior LLM responses and external state so engineers can resume from a broken step, plus Signals that cluster failure patterns across many production runs from short natural-language descriptions. The core is written in Rust and released as open source, with a managed cloud at laminar.sh. Early customers named in the launch include Browser Use and OpenHands.