Security and data handling

Your AI provider key

• Keys are encrypted with AES-256-GCM using a per-secret data key, which is itself encrypted by a master key (envelope encryption).
• Keys are never shown again after you save them, and they’re removed from logs and error messages.
• You can remove a key at any time from Connections.

Where your content goes

• Only to the AI provider you choose for that run, under your own account and that provider’s terms.
• We never send your content to a different provider. There’s no automatic fallback: if your provider is down, the run stops and tells you.
• We don’t train models on your data. Use paid-tier provider keys for client data; some free tiers let the provider use your content.
• The services that process data for us are listed on the Subprocessors page.

Nothing happens without approval

• Agents that could act outside the app (for example, creating an email draft) pause and wait for a person to approve that exact action.
• Text inside documents and web pages is treated as data, never as instructions. Agents flag content that looks like planted instructions.
• Web pages are fetched through a filter that blocks private and internal network addresses.

Accounts

• Passwords are hashed with scrypt. Two-factor authentication (authenticator app) is available in Settings.
• Roles: owner, admin, member and viewer. Billing and connections need an admin or owner.
• API keys for automation are scoped to one workspace and can be revoked.

Retention and deletion

• Uploaded files, run inputs and results are deleted after 30 days by default. Owners can change this.
• Billing records are kept so receipts stay valid.
• Owners can export or delete their workspace’s data from the Data page.

Payments

Card payments are handled by Stripe’s hosted checkout. We never see or store card numbers.

Reporting a problem

If you find a security issue, please tell us through the contact page.