Adam is not a chatbot that forgets you when the tab closes — it is a platform of persistent AI agents that remember, work in the background, and act on real tools when you tell them to. That makes it especially important that you understand exactly what data exists, why it exists, and how you stay in control of it. This policy describes the platform as it actually works.
1. What the platform does with your instructions
Each conversation is handled by an AI agent working for you. For complex work an agent can spawn sub-agents — teams of specialists that run in parallel under the same conversation and coordinate with each other. Agents also run internal processing cycles beyond the visible reply (for example organizing what they learned), and if you set up schedules or triggers, your agents wake up and work at those times even while you are away. All of this happens inside your account, under your instructions and settings, and is charged to your balance like any other activity.
2. Data the platform holds
Account
Email, name (if given), password (stored only as a salted hash) or your social sign-in identity, and — if you enable it — an encrypted two-factor secret.
Conversations and agent activity
Your messages, agent replies, the tool calls agents make, and per-turn context needed to continue the work. Sub-agent activity belongs to its conversation.
Memories
Agents maintain a persistent memory system so they get more useful over time. Memories are tiered: task-local working memory (short-lived), conversation memory (tied to one conversation), user memory (things about you and your preferences that carry across your conversations), and system-level knowledge. Each memory carries a strength that rises when it proves useful and decays when it doesn't — unused or failed memories fade and are cleaned up automatically. You can see and manage memories in the app, and they are erased with your account.
Workspaces, files, and deliveries
Agents execute work in isolated containers: files they create, code they run, media they generate, and the results they deliver back to your chat. Uploads you attach are stored so agents can use them.
Browser, voice, and vision
When an agent drives its browser you can watch the live session, and screenshots/recordings of that session are kept for you. Voice conversations are transcribed; text-to-speech output is generated from agent replies. Images and screenshots may be analyzed by vision models when a task requires seeing them.
Connectors (external services)
When you connect an external service (GitHub, Google Workspace, Slack, Jira, and others), the credential that service issues is stored encrypted with AES-256-GCM in your personal vault. It is used only for actions you or your agents initiate on your behalf, within the permissions shown when you connected. Some credentials never enter agent workspaces at all and are used exclusively through controlled internal channels (for example, Telegram delivery and billed posting to X). You can pause, disconnect, or delete any connector at any time in Settings → Connectors; deletion removes the credential immediately, and you can additionally revoke access from the provider's own security page.
Devices
You may register your own computers as execution environments. When you do, your agents can run commands on your machine, and the commands and their output are relayed through the platform to make that work. Device access is protected by a per-device credential issued at registration; you can remove a device at any time, which cuts its access immediately.
Balance and billing
Your prepaid balance and an itemized ledger of every charge — model usage, platform services (web research, media generation, voice), and connector actions that carry provider costs. Card details are handled by Stripe and never touch our servers.
Technical data
Operational logs (timestamps, request metadata, session IPs) for security and troubleshooting. There is no advertising and no cross-site tracking; the only cookies are the ones that keep you signed in.
3. What we do not do
- We do not sell your data.
- We do not use your conversations, memories, files, or connected-service data to train AI models.
- We do not read or mine your connected accounts or devices in the background — access happens only inside tasks you drive or schedules you created.
- We do not run third-party advertising or tracking.
4. Processors
Running the platform requires a small set of processors: AI model providers (which process conversation text to generate agent responses), speech providers (transcription and text-to-speech), a web-research provider, a media-generation provider, and Stripe for payments — each limited to providing its service. Your connector credentials are not shared with any of them; agents call connected services directly on your behalf.
Adam's use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements: Google user data is used only to provide the user-facing features described here, never for advertising, and is never sold.
5. Security
Secrets are encrypted at rest (AES-256-GCM under a master key); transport is TLS; internal service-to-service calls are authenticated; agent workspaces are isolated containers; and credentials are scrubbed from agent-visible output. Secret material is never written to logs in plaintext.
6. Retention and deletion
- Conversations, memories, and files persist while your account exists — persistence is the product.
- Working-tier memories expire on their own; unused memories decay and are cleaned up on schedule.
- Deleting a connector deletes its credential immediately; removing a device cuts its access immediately.
- Deleting your account deactivates it immediately (agents stop, sessions are revoked) and permanently erases your data — conversations, memories, vault, devices — after a short retention window; signing back in within the window cancels the deletion.
- Operational logs rotate on short fixed schedules.
7. Your rights
You can access, correct, export (conversation export is built in), or delete your data — most of it self-service in the app. For anything else, email [email protected]. Where statutory data-protection rights apply to you (e.g. GDPR / UK GDPR), you also have the rights of access, rectification, erasure, restriction, portability, objection, and complaint to a supervisory authority.
8. Children
Adam is not directed at children under 16, and we do not knowingly collect their data.
9. Changes
Material changes update the effective date above and are announced to active users in the app. Continued use after a change means acceptance.