Skip to content

Simulation Results

After a simulation completes, you can review the full results in the simulation history modal. Results are available for all terminal statuses — Completed, Failed, and Stopped.

  • Final screenshot — The last page the agent visited before the simulation ended.
  • Total steps taken — The number of actions the agent performed.
  • Status — Whether the simulation completed, failed, or was stopped.
  • Action log — A step-by-step record of the entire simulation.

The action log is the most detailed part of simulation results. Each entry shows:

  • What the agent saw — The page content and UI elements the agent observed.
  • What it decided — The agent’s reasoning for its next action.
  • What it did — The specific interaction (click, type, navigate, etc.).

Successful simulations automatically contribute to the agent’s knowledge graph. This means:

  • The agent gains a deeper understanding of your application’s structure and workflows.
  • Future interactions with users become more accurate and relevant.
  • Additional simulations compound — each one adds to the agent’s existing knowledge.

You can assign simulation results to multiple agents, sharing learnings across your team. This is useful when you have specialized agents that would benefit from the same application knowledge.