Skip to content

Legal

Privacy

Last updated August 21, 2026

Parallel plans and executes work on your behalf, which means it handles information about your goals, your projects and sometimes the people you contact. This page explains exactly what is collected, why, and what is deliberately never stored.

01What we collect

Only what the product needs to function. There is no advertising network and no data sale.

  • Account information: your email address, display name and authentication identifiers.
  • Profile information you provide during onboarding: occupation, interests, available time and working preferences.
  • Mission content: objectives, constraints, plans, agent configurations, deliverables and approval decisions.
  • Usage records: which Parallel Actions were consumed, by which mission and agent, and when.
  • Technical records: standard server logs, retained briefly for reliability and abuse prevention.

02What we never log

Some categories are excluded from logging by design, not by policy alone. This is a rule in the codebase, applied wherever agents run.

  • Credentials for connected services. Where real integrations exist, tokens are held encrypted and are never displayed again after saving.
  • The contents of personal documents you upload, beyond what an agent needs in the moment to produce its output.
  • Prompt text classified as sensitive by the permission layer.
  • Payment card details. Card data would be handled by a payment processor and never reaches Parallel.

03Why we process it

  • To provide the service: generating plans, agents, schedules and reports you asked for.
  • To enforce plan limits and meter Parallel Actions accurately.
  • To keep the service secure and prevent abuse.
  • To improve the product in aggregate. Mission content is not used to train third-party models.

04AI processing

Generating a plan or an agent team involves sending mission context to a language model provider. Only the information required for that specific operation is sent, and structured output is validated against a strict schema before it is stored or shown.

When no model provider is configured, Parallel runs a deterministic offline planner instead. In that mode nothing leaves the deployment, and generated output is labelled as simulated.

05Storage and retention

  • Data is stored in Postgres with row-level security: every row is owned by a user and queries execute as that user.
  • Missions you delete are soft-deleted, then permanently removed after 30 days.
  • Closing your account removes your profile, missions, agents and deliverables. Aggregated, non-identifying usage counts may be retained.

06Your rights

You can access, correct, export or delete your data at any time from Settings, or by contacting us. If you are in the EEA or the UK you also have the right to object to processing and to lodge a complaint with your supervisory authority.

07Google user data

Connecting a Google account is optional and is always initiated by you. Parallel only ever requests the narrowest scope that the connected capability needs, and each one is used for a single, stated purpose.

Parallel'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 never sold, never used for advertising, never used to train generalised machine-learning models, and is never read by a person except with your explicit permission, to resolve a support request you raised, or where the law requires it.

  • gmail.readonly: reading the messages an agent has been asked to act on, and nothing else.
  • gmail.compose: preparing drafts for your review. A draft is never sent without your approval.
  • gmail.send: sending only the specific message you approved, at the moment you approved it.
  • calendar.freebusy and calendar.events: finding free time and creating the events a mission schedules.
  • drive.file: limited to files Parallel itself creates or that you explicitly open with it. The rest of your Drive stays invisible to Parallel.
  • Disconnecting a Google account in Settings revokes the token immediately and deletes the data obtained through it.

08Sub-processors

  • Supabase: authentication, database and file storage.
  • Google: only for the accounts you choose to connect, and only for the scopes listed above.
  • A language model provider: plan, agent and report generation, when configured.
  • Vercel: application hosting.

09Contact

Questions about this page or a specific request can be sent through the contact form.

This document is written for an MVP and is not legal advice. Before Parallel processes personal data at scale it will be reviewed by a qualified adviser, and this page will be updated accordingly.

Privacy · Parallel