offload

Privacy Policy

EFFECTIVE 2026-08-13

Offload ("Offload", "we", "us") is a desktop and mobile application and cloud service that lets creators connect their platforms and tools and delegate work to an AI agent. This policy explains what information we collect, how we use it, and the choices you have. Contact: saadosman450@gmail.com.

1. Information we collect

• Account information. Your email address, used to sign in (we send a one-time code) and to contact you about the service.
• Your content. Chats, projects, drafts, and files you share with the agent. Thread event history and projects sync to our cloud backend so your work can be viewed and restored; media you attach may be stored in our cloud storage. Connected-account call arguments, results, and assistant replies derived from those results remain in the device-local turn, are not copied into cloud thread history, and are not replayed into later model requests.
• Connected account data. When you connect a third-party service (for example YouTube, TikTok, Instagram, X, Gmail, Notion, Slack, or GitHub), we access the data you authorize — such as your profile, posts, videos, messages, or analytics — solely to provide the features you use. OAuth sign-in for most tools is handled by our authentication partner (Composio), which stores and refreshes access tokens on our behalf. TikTok uses Offload's native Login Kit integration instead: Offload receives its access and refresh tokens on the server, encrypts them at rest, and never sends them to the desktop or mobile client. We do not receive or store your third-party passwords.
• Diagnostics and usage. When diagnostics are enabled for a build or runtime, crash, performance, operational error, and aggregate usage reports may be sent to Sentry and PostHog. Daemon reports are scrubbed, omit raw crash stacks, and use no persistent device identifier; their PostHog error events use a random identifier that lasts only for the running daemon process. If renderer Sentry is enabled, an application error can include stack frames, breadcrumbs, performance traces, and a masked, media-blocked replay tail. When mobile analytics is enabled, PostHog receives anonymous application install, update, open, foreground, and background events; mobile screen, touch, log, exception, survey, geolocation, and session-replay capture are disabled. Mobile Sentry can receive native and JavaScript crash reports and sampled performance traces, but mobile replay is disabled. Offload does not intentionally attach chat or connected-account content to telemetry.
• Billing and model usage. Stripe processes subscription payments and retains the billing details needed to do so. Offload records provider-reported model usage, token counts, and cost so it can show your running usage and report the billable amount to Stripe. Offload does not receive your full payment-card number.
• On-device storage. Offload's clients store session material, settings, and cached application state locally. The desktop client also stores chats, its media index, and generated descriptions locally so the agent can work next to your files. The desktop application data directory and workspace can remain after the app is uninstalled; removing the application alone does not promise to erase those files.

2. How we use information

• To provide the service: run agent tasks you request, draft and publish content you approve, sync your threads across devices, and show you your connected accounts' data inside the app.
• To process your requests with AI models: content you send to the agent (and the connected-account data a task needs) is processed by large language model providers acting as our processors. Their handling is governed by our provider configuration and agreements.
• To operate, debug, and improve Offload.
• To communicate with you about the service.

We do not sell your personal information, and we do not use your content or your connected-account data for advertising.

3. When we share information

• Service providers that host and power Offload: cloud database and sync (Convex), file storage (Cloudflare), billing (Stripe), AI model routing (OpenRouter and the model providers behind it), OAuth token management for supported non-TikTok integrations (Composio), email delivery (Resend), inbound mailboxes (AgentMail), crash reporting (Sentry), and analytics (PostHog). Each receives only what it needs to perform its function.
• The platforms you connect, when you ask the agent to act — e.g. publishing a post to your own account.
• Legal reasons, if required by law or to protect the service and its users.

4. YouTube API Services

Offload uses YouTube API Services to let you connect your YouTube channel. By using this feature you also agree to the YouTube Terms of Service, and Google's handling of your data is described in the Google Privacy Policy.

• When you connect YouTube, Offload accesses the data you authorize — your channel profile, your videos and their metadata, comments, and your channel analytics — to display it to you and to run the tasks you ask the agent to perform (for example, analyzing how your videos perform or drafting replies for your approval).
• We store imported YouTube data only as needed to provide these features and do not share it with third parties except the service providers listed above acting on our behalf. A current analytics snapshot may remain stored until it is replaced or you use the deletion controls described below.
• You can disconnect YouTube inside the app at any time. Disconnecting immediately disables Offload's local use of the connection and asks our OAuth manager to start provider-token revocation. Local disablement is recorded before that external request, so an interrupted or failed request stays disabled and can be retried. When accepted, the manager currently returns an asynchronous job receipt without an API that verifies completion, so Offload labels the revocation pending and unverified instead of claiming it has completed. Disconnect remains separate from deleting data already imported into Offload. After disconnecting, use Delete imported data in the desktop Stack screen to erase the retained imported copy. You can also revoke Offload's access directly in your Google security settings.

5. TikTok and other connected platforms

When you connect TikTok (or another platform), Offload accesses only the data you authorize on the platform's consent screen — such as your profile, your videos, and their performance metrics — to display it to you and run the tasks you request. TikTok access and refresh tokens are stored only in an encrypted server-side credential record while the connection is active. A TikTok disconnect immediately blocks local use, requests revocation directly from TikTok, and deletes the credential after TikTok acknowledges the request. If that request is temporarily unavailable, the connection remains disabled and the encrypted credential is retained only for a safe revocation retry. For other connections managed by our OAuth partner, a returned asynchronous revocation receipt is shown as pending and unverified unless authoritative completion is available. Disconnecting does not by itself erase data already imported into Offload. For supported connected tools, the desktop Stack screen offers a separate Delete imported data action after disconnect; you can also revoke access from the platform's own security settings.

6. Data retention and deletion

• Your chats, projects, and media are kept while your account is active so they can sync across your devices. On desktop, the thread action labeled Archive only hides a thread; it does not delete local or cloud history. The mobile thread screen can request deletion of that thread's cloud copy, while local Mac history and user-created backups must be deleted at their source. Removing an entire device from Offload starts a bounded cloud-mirror deletion sweep. A content-free, generation-bound deletion ledger survives device removal to stop provider backup restoration from resurrecting an erased cloud thread; it contains identifiers, high-waters, counters, status, and timestamps, not conversation payloads.
• Moving media to trash is also an archive. Empty trash starts durable permanent deletion of the selected media from Offload's local index, local content-addressed storage, cloud metadata, cloud object storage, and local generated descriptions. Local metadata is removed after the cloud accepts the deletion; background retries may continue until object-storage deletion is verified. Copies you exported elsewhere remain yours to delete.
• Delete imported data, available on the desktop Stack screen after disconnecting a supported provider, permanently removes provider metadata, analytics snapshots, schedules, and provider-operation ledgers retained by Offload. It does not remove content already published on the provider, the provider's own records, ordinary chat content, or user-created local/exported copies. Background deletion may continue after the request is accepted.
• Delete account is available in mobile Settings. After you confirm, Offload cancels active Offload subscriptions linked to the account, requests revocation of connected-account credentials, signs out every Offload session, removes the account's identifying email and login authority, and starts permanent deletion of its cloud projects, conversations, files, connected-account data, settings, and other user-authored cloud data. Deletion runs in bounded background jobs so stored objects and retries can be verified. An anonymized deletion marker, public issue attribution, and records that must be retained for legal, security, or billing purposes may remain without the account email. Local files, desktop-only archives, user-created backups, exported copies, and content already published to another platform remain at their source and must be deleted there.

7. Security

Data in transit is encrypted with TLS. Sign-in uses one-time email codes and a signed access session that can last up to 30 days. An active client may rotate that session into a new 30-day session; rotation revokes the previous session. Signing out clears the session from the current client and attempts to revoke it server-side. If that request cannot reach the service, such as while offline, the server session can remain valid until it expires even though this client no longer retains it. Offload never sees a password for you or any connected account. Access tokens for most connected platforms are held by our OAuth partner. Native TikTok access and refresh tokens are held by Offload in AES-256-GCM-encrypted server-side records, refreshed there, and retained only while needed to operate the connection or complete a requested revocation. Connected-platform tokens are never exposed to other users or shipped to the desktop or mobile client.

8. Your rights

Depending on where you live, you may have rights to access, correct, export, or delete your personal information. Email saadosman450@gmail.com and we will honor these requests. Offload is not directed to children under 13, and we do not knowingly collect their information.

9. Changes

We will post any changes to this policy on this page and update the effective date above. Material changes will be announced in the app or by email.