Privacy Policy
What we collect when you run an agent, how long we keep it, and who else sees it. Last updated June 11, 2026.
1. Information we collect
We collect user identifiers (email address, name), workspace configurations, transaction records for credit billing, and the raw inputs and outputs of task executions. Execution inputs and outputs are collected because the run cannot be performed or shown back to you on your dashboard without them.
2. How we use data
We use the collected information to:
- Authenticate logins via JWT tokens and Firebase Admin services.
- Provide and coordinate semantic-search memories for AI agents.
- Process transactions and maintain the platform credit ledger.
- Aggregate anonymous system load, error and API request latency metrics.
3. Data retention and sandbox isolation
Task execution histories and generated outputs are kept for up to 90 days, after which a MongoDB TTL index deletes them. Any file uploaded for an execution is stored inside the isolated sandbox for that run and discarded when the run completes.
4. Third-party integrations
Depending on what an agent needs, metadata may be sent to model providers (Google Gemini, OpenAI, Anthropic) or to integrations you connect (E2B Code Interpreter, Resend, Stripe, GitHub). Those transfers are covered by the privacy policy of the service receiving them.
5. Your rights
You can request access to, correction of, or permanent deletion of your account data and workspaces. Deletion can be started from the data deletion page, or by contacting us through the contact page.