Runs, Feedback & Memories
Beyond agents themselves, the Build area gives you the supporting surfaces for observing and improving agent behavior.
Runs
Runs is the history of agent executions. Use it to review what an agent did, when, and with what outcome.
- Workspace level: Build โ Runs shows runs across all agents.
- Agent level: open an agent โ Runs shows just that agent's runs.

Feedback
Feedback collects signals about how agents performed, so you can iterate on their workflows. Like Runs, it exists at both the workspace and agent level.
Datasets
Datasets are collections of data made available to agents. Manage them from Build โ Datasets.
Memories
Memories are pieces of information agents retain and reuse across interactions. Manage them from Build โ Memories.
Use these to improve agents
Treat Runs and Feedback as your feedback loop: inspect runs, read feedback, then adjust the agent's workflows in the Editor.