How AutoAGI handles your data

What AutoAGI stores

• Your account email, the profile information you choose to provide, and membership and billing references.
• Conversations, uploaded files, generated work and task activity, kept so you can return to them.
• Model usage and operating costs, recorded to enforce your weekly allowance and run the service.
• Billing references from Stripe rather than full payment-card numbers.

Account access controls separate members' conversations and files. Selected prior conversation excerpts, saved business context and knowledge files can be reused in later requests so tasks start with what AutoAGI already knows about you. Authorized service operations and support may access data when needed to operate, secure or troubleshoot the service.

Who processes your requests

To complete a request, your prompt, the relevant conversation history, saved business context and any attached or knowledge-base files may go to AI providers. Depending on the feature, that means OpenRouter and its model providers, TypeSafe for Jev, OpenAI voice services, or Harness Router agent workspaces for Cloud Harness runs.

Your data is not used to train AI models. The providers process submitted content to complete your request, under their own applicable policies. The policy asks you to avoid submitting information you are not authorized to share.

How connected apps stay safe

You sign in to each app on its own page through Composio, and your app password never enters AutoAGI. Every connection has a policy you set: fully autonomous or read only, with deletions and mass sends off until you turn them on, and an optional scope such as the repositories or channels it may touch.

Scheduled routines carry a snapshot of the connections they were saved with, and disconnecting an app removes it from every routine at once. Cloud Harness runs use your apps through the AutoAGI gateway and never hold a provider token. Every app action leaves a record of the app, the action, the resource it touched and the outcome, and you can disconnect at any time.

Agent connections

You can connect your own AI agents, such as Claude Code or Codex, to your account over MCP. Each connection is a grant you create and can change or revoke at any time: the permissions it has (delegating work, Cloud Harnesses, reading your work or your library, adding files, using connected apps, managing routines), the apps it may use at read or at read and write, a share of your weekly usage, a task limit and an expiry. How it works.

The policy you set on each connected app still applies on top of the grant, so an app set to read only stays read only for every agent, and any repository or channel scope still holds. Permissions are checked on every request. Narrowing a connection or revoking it applies on its next request. A permission you add later shows up in the agent's tool list after it reconnects or you start a new session. With write access on an app, an agent can do what that app allows there, so start apps at read only. An agent never receives a password or token for an app. The sign-in stays inside AutoAGI.

• For each connection AutoAGI stores its name, the permissions and limits you chose, the apps you shared with it and when it was last used.
• Agent keys and sign-in tokens are stored only as a hash, so AutoAGI cannot show a key again after you create it.
• The activity log holds one row for each tool call: the tool name, the time, whether it worked, a short error code, how long it took, the task id when there is one and a short app and action label. It never stores what the agent sent or received, and its rows are kept for about 30 days.
• Each app action an agent runs also leaves a record of the app, the action, the resource it touched and the outcome. The app's response is kept only when the agent sent an idempotency key to make the call retry safe, so a repeat can return the same result. That response can include content from the app, and it stays until the connection is removed, 90 days after you revoke it. A call without a key keeps only whether it worked and how large the response was.
• Anything an agent asks AutoAGI to do runs as a normal task in your account, so the prompt, the results, the files and the app actions it ran, with their results, are stored and handled as described above, including by the AI providers that process requests.
• Reading your work or your library is its own switch. Read all my work needs Delegate work. Without Read my library, the tasks an agent starts do not use your Knowledge files or profile background.

Your controls

• Edit your profile and task-email preferences in Account.
• See, change and revoke each agent connection in Account, Agent access.
• Export your account data.
• Manage or cancel your subscription through the Stripe billing portal.
• Email support@agilayer.com to request account or data deletion. Records needed for billing, security, disputes or legal obligations may be retained, and backup copies and provider records follow their own retention processes.
• Google sign-in requests only your basic identity: name, email and profile picture. It does not give AutoAGI access to your Gmail, Drive or Calendar.

What we do not claim

AutoAGI does not claim compliance certifications, regional data residency or on-premise hosting. It does not promise how third-party model providers treat submitted content. If your work needs any of those, ask before you submit it.

Questions

Sources

• AutoAGI privacy policy (updated September 30, 2026)
• AutoAGI terms