Laminar
Laminar is an open-source agent observability platform that automatically captures LLM calls, tool calls, sub-agents, costs, and tokens in readable transcript views. Its Signals feature uses a custom reinforcement-learning-trained model, flow-1, to analyze every agent run and surface hard-to-spot failures at 1/20th the cost of frontier models. The platform clusters detected issues to show the full distribution of agent failures and behaviors.
- Developer Tools
- Artificial Intelligence
- Data & Analytics
- Software Only
Funding
Raised to date
$500KRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Founder details are not available yet.
Product
Problem
AI agents execute complex, multi-step workflows involving many model calls and tool interactions, making failures difficult to pinpoint within a single trace. Teams lack an efficient way to identify unexpected failure modes at scale, and analyzing all production traces manually or with existing tools is prohibitively expensive, leaving many issues undetected.
Solution
Laminar provides an open-source agent observability platform that automatically captures LLM calls, tool calls, sub-agents, costs, and token usage, presenting them in readable transcript views. Its Signals feature analyzes every agent run using a custom model, flow-1, trained with reinforcement learning to identify hard-to-spot failures and explain why they occurred. Signal clusters reveal the full distribution of an agent's failures and behaviors, helping teams understand whether a newly reported issue has happened before. flow-1 matches frontier-model detection quality at 1/20th of the cost, making it affordable to move beyond sampling and continuously investigate all production traces.
Target Audience
Primary users are engineering teams building and shipping AI agents in production, particularly those working on software engineering, customer support, legal, and CRM workflows who need to debug, monitor, and improve agent reliability.
Features
- Automatic capture of LLM calls, tool calls, sub-agents, costs, and token usage in a readable transcript view
- Signals feature powered by flow-1, a custom model trained with reinforcement learning on trace analysis, matching GPT-6-sol detection quality at 23x lower cost per trace
- Signal clusters showing the full distribution of failures and behaviors, enabling pattern discovery and historical issue tracking
- Cost heatmaps and trace-level cost breakdowns for monitoring spend across agent runs
- Evaluation suite with datasets, labeling queues, and custom dashboards
- Browser session recording, agent debugger, and MCP access for deep debugging
- Full-text trace search, SQL editor, and custom usage limits with Slack/email alerts
- OAuth sign-in, SOC 2 Type II, HIPAA compliance, server-side PII removal, and DPA support