What does Kubit do?
Kubit is Agent Observability with Behavior Analytics. It labels and scores every span of your agents’ OpenTelemetry traces with Polaris, groups agent runs by execution path in Flows, and runs funnels, retention and cohorts on agent traces and clickstream events, separately or joined. See What is Kubit.Who is Kubit for?
Kubit is for AI product engineers who build and run LLM agents in production. Product managers and analysts use it to see which agent behavior leads to a completed task, a support ticket or a cancellation.Is Kubit product analytics or agent observability?
Kubit is one product. Product Analytics runs reports on any behavior events, including agent traces and clickstream events. Agent Observability is the set of features that only work on traces: Polaris, Flows and Triggers.What is Polaris?
Polaris is Kubit’s in-house small language model. It labels every span with an annotation type (Action, Exception, Misstep, Warning or Behavior), scores conversations against your own questions and can be fine-tuned on the corrections your team makes. It labels a span in under a second, with no additional LLM cost or sampling. See Polaris.What is a Misstep?
A Misstep is a call that ran, often with an OK status, but did the wrong thing, such as calling the wrong tool or passing the wrong parameter. An Exception shows in your error logs. A Misstep usually doesn’t. See Find Missteps that return OK.Does Kubit sample traces?
Polaris labels every span, not a sample. Flows needs every run labeled to show every execution path with its count, and Triggers reads the labels. See Polaris.Which warehouses does Kubit work with?
Kubit queries data in place in Snowflake, BigQuery, Databricks and ClickHouse. Polaris labels and scores can be stored in all four. Redshift is listed in the app but not supported yet. See Bring your own warehouse.Where does Kubit store my trace data?
When your traces are in your warehouse, Kubit queries them there and writes Polaris labels and scores to a schema you pick in the same warehouse, in tables that start withKUBIT_. No data silo to secure. Traces retained in your control. See Where Polaris results are stored.
How do I send traces to Kubit?
There are three ways: the agent plugin for Claude Code or Cursor, the Kubit SDK (kubit-otel for Python, @kubit-ai/otel for Node), or the stock OpenTelemetry exporter. If your traces are already in Databricks, Snowflake, BigQuery or ClickHouse, Kubit queries them there instead. See the Quickstart.
Can I use Kubit with Langfuse or Braintrust?
Yes. The agent plugin addsKubitSpanProcessor next to your Langfuse or Braintrust processor, and your current tool keeps receiving every span. Self-hosted Langfuse traces in your ClickHouse can be read in place. See Use Kubit with Langfuse.
Can I try Kubit without changing my app?
Yes. Create a Sandbox workspace and upload a trace file in OTLP, Langfuse, Phoenix, MLflow, LangSmith, Braintrust, Datadog, AWS CloudWatch or Snowflake Cortex format. See Logfire traces in a Sandbox for a worked example.How do I get access?
Agent Observability is in early access. Join the waitlist.Related
- Glossary: every term on this page, defined.
- What is Kubit: the product in one page.