Trajectories
See user behavior in your app before, during, and after an agent conversation.
Open your coding agent in your project's folder, or in an empty folder for a new app.This runs in a coding agent on your computer.
Overview#
A Trajectory records a session in your app. It can include clicks, form edits, page changes, and network requests.
CopilotKit Intelligence stores these events and links related agent conversations. You can inspect what a person did before, during, and after a conversation.
Early access with our engineering team
Trajectories are in early access for proofs of concept with our engineering team. If you need help with access, book time with our engineers.
Start with the quickstart to record your first session in self-hosted Intelligence.
Choose a guide#
| Guide | What you will do |
|---|---|
| Quickstart | Run Intelligence locally and find your first saved Trajectory. |
| Capture interactions | Add capture to an app that uses CopilotKit. |
| Captured data | Review the recorded fields, redaction limits, and controls that exclude data. |
| API reference | Look up learning options, Core methods, and capture error codes. |
How Trajectories relate to Automatic Learning#
Trajectories provide evidence about how people use your app. Automatic Learning analyzes evidence and proposes Skills for you to review.
Capture alone does not start an analysis or change an agent. To use a Trajectory as evidence, add it to a Learning Space. See Assign browser activity to a Learning Space.
After you review and publish a Skill, Skill delivery makes it available to an agent.
User Memory stores facts about a person across conversations. It is a separate feature. You do not need User Memory or Automatic Learning to record and inspect a Trajectory.