Data Inventory
Status: Published · Last reviewed: 2026-10-22 · Next review: 2027-04-16 Regulatory reference: Article 30 GDPR.
This inventory lists the categories of personal data Syncflow processes. It is written at category level so that users can understand what we hold without exposing the underlying technical design. Specific fields within each category are handled in accordance with the controls summarised in the DPIA and the Tier Classification.
1. Account identity
What it is. Information that identifies the account holder: the email address used to sign in, and — where the user chooses to provide them — a display name, a country, and a profile picture.
Source. Provided by the user during sign-up or inherited from the identity provider the user chose to sign in with.
Why we hold it. To create the account, authenticate the user, communicate service messages, and personalise the interface.
2. Authentication credentials
What it is. The cryptographic and session material that proves the user's identity when they sign in. This includes:
- Third-party sign-in linkage where the user signs in through an identity provider (e.g. GitHub, Google);
- Active session artefacts created when the user signs in;
- Short-lived one-time tokens issued during email-based verification; and
- Passkey (WebAuthn) registrations where the user has enrolled one.
Source. Generated by Syncflow or by the identity provider at sign-in time; the user does not supply this data directly.
Why we hold it. To authenticate the user and keep them signed in.
3. Account preferences and configuration
What it is. The user's time-zone and working-hours preferences, their notification preferences, their onboarding state, and any optional answers given during onboarding.
Source. Provided by the user through the settings interface.
Why we hold it. To deliver the scheduling and reminder features, and to tailor the interface to the user.
4. Subscription and billing references
What it is. References that identify the user's account inside Syncflow's payment processor. Card details and full billing addresses are collected and stored directly by the payment processor — Syncflow never receives or stores them.
Source. Issued by the payment processor at the time of subscription.
Why we hold it. To check the user's subscription status and to support billing queries.
5. User-generated task content
What it is. The tasks, breakdowns ("crumbs"), reusable templates, export templates, comments, and tags that the user creates inside Syncflow. Free-text fields within this category are written entirely by the user.
Source. Supplied by the user; some crumb content is generated by the AI decomposition feature when the user asks for it.
Why we hold it. This is the core product; without it Syncflow cannot function.
Anonymisation. Once a plan is finished or archived and has been idle for 90 days, its text is replaced and only its structure and numbers are kept, unless the user chose to keep it in full. See Tier Classification §"Anonymisation of idle plans".
Deletion and export. A deleted task or crumb spends 30 days in the trash, hidden from every read and restorable by the user, before a daily job erases it; the user can erase it sooner, or everything at once. The whole category can be downloaded as JSON or CSV at any time, trash included. See Tier Classification §"The trash" and §"Data export".
6. Progress and scheduling telemetry
What it is. Information recorded automatically as the user works through their tasks: which crumbs have been started, completed, or skipped, and when; user-recorded actual vs. estimated effort; streak and progress counters.
Source. Generated by the Syncflow application as the user interacts with their own data.
Why we hold it. To surface progress back to the user, to keep streaks and reminders accurate, and to improve future effort estimates for the same user.
7. User-configured integrations
What it is. Optional integrations the user sets up themselves — for example, a user-supplied third-party AI-provider key. Credentials supplied by the user are held encrypted at rest.
Source. Provided by the user when they configure the integration.
Why we hold it. Only to operate the integration the user has requested. The user may revoke the integration at any time.
8. Support and correspondence
What it is. Messages the user sends to Syncflow through support channels, and our replies.
Source. Provided by the user.
Why we hold it. To answer the user's support request and to keep a record of what was agreed.
9. Operational metadata
What it is. Routine service metadata — authentication events, service-availability telemetry, error diagnostics. Request contents (including task content) are excluded from diagnostic telemetry by design.
Source. Generated by the service.
Why we hold it. To operate the service safely, diagnose faults, and respond to security incidents.
In-app funnel events (iOS). The iOS app reports a short list of named moments (for example, that the upgrade screen was shown and why, or that a purchase started or succeeded), with a few low-detail labels and no task content. They are kept against the account for 30 days so that a view can be matched to a purchase, then folded into daily aggregate counts that carry no account identifier and include differential-privacy noise, and the per-account rows are deleted. These are operational data: the app has no analytics switch for them. The web app's own product events are not in this list, because the web app sends them only if the user accepted analytics (§9a).
Signup source. When an account is created from a link that names a campaign, we keep the campaign's name, the website the link was clicked on, and the time, on that account only. It is written once, within two hours of signup, and never for an existing account. It is carried in the link itself, not in a cookie or in browser storage. We use it to know which channel brought an account. It holds no personal data beyond the account it belongs to and is deleted with the account.
9a. Product analytics (consent only)
What it is. Collected only if the user accepts analytics in the cookie banner: a random browser identifier and, once signed in, the Syncflow account identifier; named product events with a few low-detail labels (no task or crumb content); checkout start and completion; browser error reports; and session replays in which every input and all on-screen text is masked. IP addresses are discarded. Visit measurement (page views and referrers) is collected by a second service, Ahrefs Web Analytics, under the same consent.
The named product events include the ones the web app sends to Syncflow's own servers (for example, that a Help me start session was started, was blocked because a key or plan limit was reached, or was stopped). The web app queues them only while the choice is yes, and drops any not yet sent the moment the user says no. Once received they are kept against the account for 30 days.
Source. The user's browser, and the server for checkout events.
Why we hold it. To learn which features are used, where people leave the path to a paid plan, and which errors occur.
Retention. Up to one year, and deleted sooner on request. The browser cookie and storage last up to six months and are deleted when the user withdraws consent. The consent choice itself is stored in the sf_consent cookie for six months.
Recipients. The analytics processors in the Processor Data Flows §2.8 and §2.9.
9b. Measurement without a choice (Vercel Web Analytics)
What it is. Page views, and a few named actions (for example, that a plan was created, a step was marked done, or the upgrade sheet was opened) with a few low-detail labels, counted by Vercel Web Analytics, which belongs to our hosting provider. It sets no cookie, uses no browser storage and is not keyed to the Syncflow account, so it is measured whether or not the user accepted analytics. The address it receives has the query string and any task, rule set or crumb id removed before it is sent. No task or crumb content is sent.
Source. The user's browser.
Why we hold it. To see which pages are visited and where people leave the path to a paid plan, without following anyone.
Retention. Page-view counts are kept according to the provider's published retention schedule; nothing is stored against the Syncflow account.
Recipients. The hosting provider, in Processor Data Flows §2.2.
10. Autopilot automation data
What it is. Data created by Autopilot, the feature that lets a user write rules ("when a crumb is due today, draft it") and have Syncflow act on their own tasks for them. It has six parts:
- Rule sets. The rules the user built: the conditions, the actions, the limits they set, and the plain-English explanation of the rule. This is user configuration, not content about the user.
- The run ledger. One row for every decision Autopilot made, whether or not it acted. Each row records which rule fired, on which task or crumb, what was decided, why it was skipped or downgraded, which AI provider and model were used, and how many tokens the request cost. It also stores a snapshot of the facts the decision was based on, so that the question "why did my rule do this" has an answer later. Those facts include the user's own task and crumb text as it stood at that moment.
- Artifacts. The output an action produced: research notes, a plan, a draft, a result from the user's own machine. This is AI-generated content derived from the user's task text, held separately from the task so that nothing the user wrote is ever overwritten.
- Assignments. Work briefs queued for a local executor (see Processor Data Flows §2.7). A brief holds the task and crumb text, the acceptance criteria, and the user's instructions, and nothing else.
- Executor registrations. A record of each local client the user has connected: a client identifier generated on that machine, a name the user gave it, its kind, its declared capabilities, and the time it last checked in. No content.
- The automation profile and limits. The role and self-declared areas of expertise the user entered so that rules can be written about them, their daily capacity, their guardrail settings, and their pause switch. Self-declared expertise is an occupational preference, not special-category data within the meaning of Art. 9.
Autopilot also records crumb expansions: when a crumb is split into smaller ones, a snapshot of the crumb as it was before the split is kept so that the resulting items can still be explained.
Source. Rule sets, the profile and the limits are provided by the user. The ledger, artifacts, assignments and executor registrations are generated by the service while carrying out the rules the user configured.
Why we hold it. To run the automation the user asked for, to show the user what was done on their behalf and why, and to keep that record long enough to investigate a complaint or a security incident.
Tier and retention. The same tier as the task content it is derived from. Retention is not open-ended: see the Tier Classification.
Categories of personal data that Syncflow does NOT process
- Special-category data (Art. 9 GDPR): racial or ethnic origin, political opinions, religious or philosophical beliefs, trade-union membership, genetic data, biometric data (other than the WebAuthn passkey mechanism, which uses cryptographic keys rather than biometric templates — raw biometrics never leave the user's device), health data, sex life, or sexual orientation.
- Children's data (Art. 8 GDPR): Syncflow is not directed at children and does not knowingly process data from users under 16.
- Criminal-offence data (Art. 10 GDPR): not processed.
- Location data: Syncflow does not collect precise device location.
- Marketing profiles: Syncflow does not build advertising or behavioural profiles of users.
A note on free-text fields
Free-text task and crumb content may, at the user's own volition, contain information that is personal to the user or to third parties. Where Autopilot is used, that same text is copied into the fact snapshot on a run and into any artifact or brief derived from it, so it exists in more than one place until the retention period for those records expires. Users are asked in the Privacy Policy not to place special-category, confidential, or third-party-identifying information into task content. The risk arising from this is assessed in the DPIA.
Storage location
Personal data is stored within Syncflow's managed database and, for profile pictures only, within a managed object-storage service. See the Processor Data Flows for a summary of who receives what and where it is processed.