Laminar trained an AI model to catch silent agent failures at 23x lower cost than GPT-6-sol

Launch HN: Laminar (YC S24) – We trained a model to catch silent agent failures

Laminar, an open-source observability platform for AI agents, has released flow-1, a model trained with reinforcement learning to detect errors in agent traces. On their benchmark, it matches GPT-6-sol in detection quality while being 23 times cheaper, making it affordable to analyze every production trace without sampling. The model runs inside a harness called Signals that treats a trace like a code repository, grepping through spans and returning findings as customizable JSON events. Similar events are automatically clustered to surface recurring problems.