Privacy Policy

Fabian Schucht is the controller for account administration, the public website, direct customer relationships, security, product analytics, and support. When a business customer submits personal data for Task Machine to process on its behalf, that customer is normally the controller and Task Machine acts as its processor under the Data Processing Addendum.

Fabian Schucht
C. General Vives 1 6E
35006 Las Palmas de Gran Canaria
Spain
NIE: Z1096165J

We collect information you provide directly when you create an account, join a workspace, submit content, configure a connector, contact support, make a commercial inquiry, or manage a subscription. A workspace customer or another member may also provide your information when they invite you or submit content about people connected with their work.

Connected services can return identity, repository, communication, file, public-web, and tool-result data according to the connection and instructions you or your workspace selected. Google can provide account information when you choose Google authentication. Link and Stripe provide transaction status and limited billing records without giving Task Machine complete card details.

We collect website, device, usage, and security information automatically when browsers, local daemons, managed workers, and connected services communicate with Task Machine. The categories and limits of that collection are described above and in the Cookie Policy.

When you connect Google Workspace, agents can access the data allowed by that Connector's permissions to carry out work you or your workspace request. Depending on the integration, this can include files, spreadsheets, documents, slides, email, calendar items, contacts, chat messages, directory information, and search results. This content can contain personal or confidential information. Permissions determine whether an integration can read, search, create, or change data.

Relevant Google content can be included in prompts, tool results, model responses, and resulting work. It may be shared with OpenRouter and the selected downstream model provider for Task Machine-managed AI, with customer-configured providers or external tools you choose, and with authorized workspace members through product records. Curated Google Workspace Connectors explain this in their Sign-in step before account consent. View the Subprocessor Schedule.

Task Machine does not use Google Workspace data for advertising or to train its own general-purpose models. Managed model requests require OpenRouter's no-training provider routing, as described below. This routing requirement does not govern customer-configured model providers or external services, which remain subject to their own terms and settings.

Retrieved content and results can remain in Tasks, Chats, documents, and Run history under the retention and deletion rules in this policy. Disconnecting a Connector or revoking Google access does not delete existing product history. You can submit a privacy request through Privacy & data or contact us by email; we assess the records, your authority, and applicable legal obligations. Disconnecting is not a promise that copies already sent to a provider have been deleted.

Google's additional rules for access to Google user data are published in the Google API Services User Data Policy.

We use personal data to create and secure accounts, operate workspaces, execute and resume agent work, keep browser-use visual audit history, deliver model and connector requests, keep run history, provide support, process subscriptions and refunds, prevent abuse, diagnose failures, improve the product, communicate service information, and meet legal obligations.

We rely on performance of a contract and steps requested before a contract to provide accounts, workspaces, support, and paid service. We rely on legitimate interests to secure and improve Task Machine, prevent abuse, support customers, keep appropriate business records, and measure public reach, after considering the effect on the people involved. We rely on legal obligations for tax, accounting, regulatory, and lawful-request records. We rely on consent for optional persistent public-website analytics cookies, product update emails, and other optional communications where consent is required.

You can withdraw consent without affecting earlier lawful processing. You may object to processing based on legitimate interests, and we will assess the request against any compelling legitimate grounds or legal requirements that apply.

Cookieless DataFast analytics runs on the public website by default. Before you accept analytics, and after rejection, DataFast can process page paths, referrers, campaign parameters, IP address, and user agent, use session-only browser storage, and derive a server-side pseudonymous visitor identifier with an approximately daily rotating salt. This mode does not link visitors across taskmachine.io and iterationlayer.com or forward identifiers to checkout. After you choose Accept, DataFast can additionally use persistent first-party visitor and session identifiers, briefly carry pseudonymous journey parameters between those domains, and connect that journey with Stripe revenue attribution. We do not use DataFast for advertising or unrelated cross-site tracking.

Signed-in product activity is measured server-side through PostHog Cloud (EU). Events contain identifiers, enumerations, counts, durations, routes, and cost figures rather than prompts, message bodies, file contents, names, email addresses, or raw errors. PostHog receives an IP address only for user-originated events where geolocation is used. No PostHog browser SDK, analytics cookie, session replay, or public website page-view tracking is enabled.

The Cookie Policy lists each browser cookie and explains how to change your choice. Read the Cookie Policy .

We do not sell personal data. We do not use personal data for targeted advertising or unrelated cross-site tracking. If we create aggregated or de-identified information for service analysis, we use it only where people and customers are not reasonably identifiable and do not attempt to re-identify it.

We use Scaleway for hosting, managed PostgreSQL, object storage, transactional email, observability, and backups. Blaxel processes managed sandbox and worktree data. PostHog Cloud (EU) provides product analytics, DataFast provides cookieless public website analytics with optional consent-based persistent attribution, Cloudflare provides authoritative DNS, application proxying, web application firewall protection, TLS termination, and Turnstile abuse prevention, and OpenRouter and the selected downstream model provider process managed model requests. Managed structured evaluations also use OpenRouter and may include bounded feature state and typed questions. Customer-selected connectors, including Exa when you install or invoke it, and local coding tools receive data only when you or your workspace configure or invoke them.

Link is the merchant of record for Managed Payments and processes payment, tax, fraud, refund, and dispute data as an independent controller under its own terms. Authentication services and customer-selected coding tools, model providers, MCP servers, and connectors may also act under their own privacy terms when you or your workspace choose them.

We may disclose limited information to professional advisers, insurers, auditors, public authorities, courts, or other parties when reasonably necessary for advice, legal compliance, security, claims, or the protection of rights. Information may also be disclosed as part of a proposed or completed corporate transaction, financing, reorganization, or transfer of the business, subject to appropriate confidentiality and data-protection safeguards.

View the Subprocessor Schedule or read the Data Processing Addendum.

Account, workspace, task, chat, transcript, browser-use action history, and related product records are normally kept while the account or workspace remains active and afterward only for as long as needed for service operation, security, disputes, legal obligations, or an agreed customer instruction. Browser-use records and remaining private frames follow their owning Task or Chat and do not have a separate automatic expiry. Disabling future browser-use screenshots retains existing images unless the owner or admin chooses irreversible deletion; that deletion removes matching active-storage images while preserving action logs, and residual backup copies expire through applicable backup cycles. Historical waitlist contact details remain until the launch announcement and related follow-up are complete or the contact asks us to delete them, subject to applicable legal obligations. A public inquiry Task and its submitted contact details remain under that operator project lifecycle until they are no longer needed for inquiry follow-up, legal obligations, or a related contract. Notification delivery records and their link-open times follow the workspace records they belong to, and product update email and Morning briefing records are deleted with the account. Archiving a workspace or record does not delete its history. Signed-in users can submit a privacy request from the Privacy and data page; otherwise, contact us by email. We assess each request's records, scope, and legal obligations.

Scaleway Cockpit application logs are configured for 30 days and metrics for 15 days. Deleted Vault credentials remain recoverable from trash for 30 days before permanent purge.

Managed worktree files have a separate lifecycle from Tasks, Chats and Teams. Task and Chat files are normally scheduled for deletion seven days after qualifying activity, on the date shown in the product. Qualifying activity includes persisted Task changes, comments, attachments, decisions, Chat turns and responses, and completed Runs. Team check-in files are normally scheduled for deletion seven days after storage becomes ready. Automatic check-ins do not extend that period. Completing, canceling, or archiving a Task starts a separate 24-hour recovery period that may schedule the files for earlier deletion.

Authorized workspace managers can add one, seven, or fourteen days to the ordinary deadline or request immediate deletion when no active Cloud execution still uses the files. Loss of workspace access gives active-Task files a seven-day return window. Security, account blocking, explicit deletion, and other mandatory cleanup may require earlier removal. Deleting managed worktree files does not delete the Task, Chat or Team, transcripts, comments, receipts, or separately saved results. Local worktree files remain on the customer's machine and are not deleted by Task Machine.

Provider records and backup copies follow the applicable provider settings and backup cycles. Cloudflare caches and security records have separate lifecycles from our Scaleway records. Cloudflare's published policies retain personal data according to service purposes and legal obligations; our application log retention does not set Cloudflare's retention. When primary data is deleted, residual backup copies are removed as those backup cycles expire unless preservation is required for security, legal claims, tax, accounting, fraud prevention, or another legal obligation. Analytics records follow the configured vendor retention settings.

Task Machine-managed structured evaluations route through OpenRouter with a per-request zero-data-retention requirement. This does not change Task Machine's own retention of product records and evaluation results.

Every Task Machine-managed AI request through OpenRouter requires its no-training provider routing. Provider fallback cannot remove that requirement, and callers cannot turn it off. If no eligible provider can serve the selected model, the request fails rather than relaxing the requirement. This does not mean zero data retention or delete Task Machine's stored records. Customer-configured model providers and external tools are not covered by this managed routing requirement.

Workspace managers can require EU processing for managed AI in Privacy & data. When enabled, subsequent Task Machine-managed AI requests for that workspace use OpenRouter's EU API and qualifying model-provider endpoints in the European Union or EEA. This includes later requests in ongoing work. Requests already sent may finish. If no qualifying endpoint is available, the request is blocked rather than routed globally.

OpenRouter's EU-Only Processing Addendum to our DPA covers API payloads, including prompts, inputs, outputs, and other content sent through the EU API, subject to its applicable-law and written-instruction exceptions. It does not restrict account, billing, authentication, or other non-payload operational records. The workspace setting does not restrict other Task Machine services, application storage, managed worktree infrastructure, external tools, Connectors, or customer-owned model providers used by Local Workers. Task Machine-managed workspace AI features remain covered even when work runs on a Local Worker.

EU processing does not itself mean zero data retention. The workspace zero-data-retention requirement is separate; when both requirements are enabled, managed requests use only endpoints that qualify for both. Neither control changes Task Machine's own retention of workspace records. AI features that cannot meet the workspace requirements are unavailable while those requirements are enabled.

Task Machine is operated from Spain. We use service providers to host Task Machine, run managed Cloud work, deliver network traffic, analyze product use, process payments, and provide model services. The Subprocessor Schedule identifies those providers and the location information available for each service.

For managed Cloud execution, Task Machine requests Frankfurt, Germany, as the infrastructure region for worktree volumes and attached sandboxes. This describes the region requested from and reported by the infrastructure provider. It does not mean that all related processing, network routing, logs, backups, support access, or model-provider processing occurs only in Germany or the EEA.

Some providers operate globally and may process personal data outside the EEA. Cloudflare, for example, uses a global network, including in the United States. Other providers and selected model developers may also process data outside the EEA. Managed AI requests for workspaces requiring EU processing are limited to qualifying regional endpoints as described above. When a transfer requires a safeguard, we rely on an adequacy decision, the European Commission Standard Contractual Clauses, or another lawful transfer mechanism and assess supplementary safeguards. Cloudflare's Data Processing Addendum includes Standard Contractual Clauses for restricted transfers.

Depending on applicable law, you may request access, rectification, deletion, restriction, portability, or objection. You may withdraw consent where processing relies on consent. Signed-in users can submit a request through the Privacy and data page, while anyone may contact us by email. We may ask for information needed to verify your identity, authority, and request scope. An authorized representative may submit a request where applicable, but we may require proof of that authority and verify the request with you.

A workspace customer may need to handle requests concerning personal data it controls, and Task Machine will assist that customer as described in the Data Processing Addendum. You may complain to the Agencia Española de Protección de Datos or another competent supervisory authority in your place of residence, work, or the alleged infringement. Task Machine does not use personal data to make solely automated decisions that produce legal or similarly significant effects about you.

We may update this policy as Task Machine, its providers, or legal requirements change. Material changes receive a new version and an appropriate notice or renewed acknowledgement where required.

For privacy requests, email [email protected] .