The archive · AI & Models · Product decision · 2024–2026
Laminar: agent observability that debugs 40-minute runs wins Browser Use, lands $3M seed
YC S24 founders built open-source Rust observability for long-running agents — one-line tracing, replay-debug, failure patterns; $3M seed from Atlantic.vc
Laminar (lmnr-ai)
What the business is
Open-source (Apache-2.0, Rust) observability platform for AI agents: one-line tracing of every LLM call and tool use, browser-session replay synced to traces, a debugger that re-runs from any step by mocking prior LLM calls, and AI 'Signals' that surface failure patterns.
Starting capital：$3M seed announced 2026-03-17, led by Atlantic.vc with Y Combinator and AAL.vc, plus angels Ben Sigelman (OpenTelemetry co-creator) and Ant Wilson (Supabase CTO)
How it started
Robert Kim (infrastructure at Palantir and Bloomberg) and Dinmukhamed Mailibay (payment infrastructure at AWS) — childhood friends from Kazakhstan who studied at KAIST — founded Laminar after YC's Summer 2024 batch, where the entire observability stack was built for single LLM calls and simple chains, not agents that run for 40 minutes and make hundreds of decisions.
What happened
Launched in 2025, Laminar signed Browser Use, OpenHands, Rye.com, and Alai as customers; its SDK is integrated into OpenHands' agent and benchmarking infrastructure and it is the default observability solution in Browser Use's documentation. On 2026-03-17 it announced a $3M seed led by Atlantic.vc (with YC, AAL.vc, and angels Ben Sigelman and Ant Wilson); the YC Launch post on 2026-04-21 detailed the one-line trace, replay-debugger that mocks prior LLM calls, and Signals pattern detection.
How it ended up
Still live and scaling: dev.co's profile (2026-07-08) shows lmnr-ai/lmnr at ~3.1k GitHub stars and 216 forks, with a v0.2.0 release on 2026-06-18 and a managed platform at laminar.sh alongside self-hosting.
Background
Laminar is an open-source observability platform for AI agents, founded by Robert Kim and Dinmukhamed Mailibay — YC S24 alumni who met growing up in Kazakhstan, studied at KAIST, and worked at Palantir/Bloomberg and AWS. Their premise: tools built for single LLM calls and simple chains can't debug agents that run for 40 minutes and make hundreds of decisions.
The product captures every LLM call, tool use, and function execution with one line of code; for browser agents it records full browser sessions synced to traces, so developers see what the agent saw. Its Agent Debugger re-runs a failed agent from any step by mocking all prior LLM calls — restoring conversation history and external state without spending tokens — and Signals uses AI to cluster failure patterns across thousands of runs.
Since launching in 2025, Laminar has counted Browser Use, OpenHands, Rye.com, and Alai among its customers; its SDK is integrated into OpenHands' infrastructure and it is the default observability solution in Browser Use's documentation. On 2026-03-17 it announced a $3M seed led by Atlantic.vc with YC and AAL.vc, plus angels Ben Sigelman (OpenTelemetry co-creator) and Ant Wilson (Supabase CTO).
As of 2026-07-08 the open-source repo lmnr-ai/lmnr sits at ~3.1k GitHub stars and 216 forks, with v0.2.0 released 2026-06-18; the company runs a managed platform at laminar.sh alongside self-hosting, positioning agent-native observability as critical infrastructure as agents move from prototypes to production.
What has to be true
- The founders lived the pain — agents failing irreproducibly — so the replay-from-the-failing-step debugger came from their own workflow, not a feature list.
- One-line integration and auto-patching of popular frameworks (Claude Agent SDK, AI SDK, LiteLLM, Browser Use, Stagehand, OpenHands) removed the adoption tax that kills developer tools.
- Being the default observability in Browser Use's docs and inside OpenHands gave distribution through the frameworks themselves, not just sales.
- Open source under Apache-2.0 with self-hosting answered the trust and data-sovereignty questions that enterprise teams ask before wiring a tool into their agents.
- The $3M seed — with investors like OpenTelemetry's co-creator — validated the category at a moment when at least one well-funded AI company had built similar capability in-house.
What can be applied
When existing tools can't handle your users' new problem, build the tool you needed, open-source it so the ecosystem adopts it as default, then raise — adoption proves the category.
Aftermath
As of 2026-07-08, Laminar is live and scaling: ~3.1k GitHub stars, 216 forks, a v0.2.0 release (2026-06-18), and a managed platform at laminar.sh alongside Docker-based self-hosting. The seed funds product development and go-to-market as agents move from prototypes into production, with Signals — AI detection of failure patterns across runs — the differentiator customers cite over incumbent observability platforms.
Sources
- Agent debugging startup Laminar raises $3M seed to tackle the observability gap in AI agents
- Launch YC: Laminar – Understand why your agent failed. Iterate fast to fix it.
- Laminar: Open-Source AI Agent Observability Platform
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card