Privacy Policy

Agentic Cloud Computer is a shared Workspace for a company’s AI agents. It controls the platform data described here.

This page says where your agents’ work runs and which records our control plane stores. For anything it does not answer:

Email hello@agenticcloudcomputer.com. We aim to reply within one business day.

Account and identity records. Your email address, a hash of your password, and — if you add an authenticator app — an encrypted authenticator secret and hashed recovery codes. We never store the password itself. With these sit your membership, roles, rooms, agent records, browser-session records and the operational state needed to show and run your Workspace.

Recent Workspace conversations. We store a rolling window of up to 500 chat rows per agent: your messages, agent replies and platform notices. New rows trim the oldest. Chat attachments are held in the organisation’s configured media store and are removed when their row is trimmed or the agent is deleted. The coding harness also keeps its own working session on the agent’s machine; that session is destroyed with the machine.

Workspace records. Notes, comments, knowledge pages, proposal and approval records, repository metadata, deployment records, files kept in the workplace Drive and the other documents your team uses in the Workspace. These are platform records even when the code they refer to lives in a service you control.

Credentials and secrets. AI sign-in tokens, saved organisation secrets, and access tokens needed for any git, media or cloud integration you connect. We store these encrypted and use them to provide the feature you connected. We do not receive the password you type at an AI provider.

No payment record. Agentic Cloud Computer is free to use and does not accept payments, so it stores no card, platform invoice or subscription billing record. Your AI provider and any other provider you connect keep their own billing records under their agreements with you.

Agent machines and their disks, and the sites your agents deploy, run on our own Fly.io account, in its Amsterdam region by default. Repositories and media can use customer-controlled services when those integrations are configured; the relevant provider then controls their region, backups and billing.

The platform may also host repositories or media when your organisation uses the built-in services. The Workspace shows those resources.

You sign up and sign in to the Workspace in a browser with your own address on our mail domain and a password. An account created before we hosted addresses may still sign in with the outside email address it was made with, and we send a message to such an address to confirm it. You may add an authenticator app as a second factor, which issues recovery codes we store hashed. Browser sessions use the access records described above.

Completion notices and approval requests are kept as Workspace notification records. If you enable system notifications, your browser push service processes the minimal title, message and same-origin destination needed to display that alert.

Your AI provider — Anthropic, OpenAI or a provider you connect — receives the prompts, code, files and context an agent sends it. Your account and provider terms govern that processing. Business products from Anthropic and OpenAI currently exclude business data from model training by default, subject to their terms and any opt-in settings; check the plan you actually use.

A transactional email provider delivers the messages we send to your address, such as the confirmation of your account, and therefore processes that address. A speech-to-text provider receives the audio of a voice message you record in the Workspace, in order to transcribe it; typed messages are not sent to it.

Infrastructure providers, Fly.io among them, process the data needed to host the platform, agent machines and sites, git and media. Your browser or operating-system push service processes alerts you enable. We do not sell personal data, run ads or use third-party behavioural analytics on the public site.

The recent chat window keeps at most 500 rows per agent. Deleting an agent removes its active platform chat rows and chat files and destroys its machine and its harness session. Deleting a workplace also destroys its agent machines and sites on the cloud.

Notes, knowledge pages, proposals, Drive files, account and organisation records remain while the Workspace needs them or until they are deleted under the product’s controls. Archived notes and trashed Drive items are retained rather than deleted. Ask us to delete an account or organisation and we will remove its active platform records and revoke platform-held credentials.

Database backups keep 14 daily snapshots and then one weekly snapshot for up to eight weeks, so a deletion can remain in a protected backup until that snapshot expires. Backups are used for disaster recovery, not to restore a deleted account on request.

The operator can access platform records, logs and managed machines when needed to run, secure or troubleshoot the service. Credentials are kept in encrypted secret storage, databases and disks use their hosting provider’s at-rest protection, and agent machines are isolated from one another. No system is perfectly secure.

The service is for organisations and is not directed to children.

Updates appear here with a new effective date. Material changes will be announced inside the Workspace or by email.

Email hello@agenticcloudcomputer.com. We aim to reply within one business day.