LangSmith
Trace AI agents and inspect quality, cost, and latency
Analyze AI agent behavior alongside data in your warehouse

Kubit offers behavior analytics for AI agents, grouping execution traces by the paths they follow. Its agent observability product labels spans and displays flows as Sankey diagrams so teams can investigate retries, failures, latency, and cost across many runs. Funnels and cohorts help compare patterns instead of opening each trace separately.
The platform is designed to query traces in supported data warehouses and join them with business activity, experiments, or campaign data. Its current site lists Databricks, Snowflake, BigQuery, and ClickHouse integrations, along with trace-file uploads for trying the analysis. The agent observability offering is available through an early-access program, with additional capabilities shown on its roadmap.
Tried it? Share a useful thought or ask a question.
Good conversations start with a hello.
Sign in to comment