Privacy Policy
No drafted text is published here on purpose. A policy that looks reviewed but is not is worse than an absent one — it implies a commitment nobody made. Reviewed text has not been published for this document yet; it will appear here once it has been.
What this system actually holds
Engineering fact, not legal advice — the inventory counsel would otherwise have to extract by interview.Account data
Examples: Email address, display name, Argon2id password hash, session records with IP address and user agent, authentication event log.
Source: Provided by the user or by an operator creating the account.
Retention: For the life of the account. Sessions are purged 30 days after expiry; auth events are currently retained indefinitely.
For counsel: Contract performance — for counsel to confirm.
Customer competitive analysis
Examples: Competitor names, URLs, cluster taxonomy, watch priorities, battlecards, analyst judgements.
Source: Authored by the customer.
Retention: Until the customer deletes the workspace, which cascades to every child row.
For counsel: Contract performance — for counsel to confirm.
Third-party personal data in collected signals
Examples: Names, employers and job moves of identifiable people appearing in news headlines and captured page text. The live corpus contains exec-hire signals naming individuals.
Source: Public news feeds and competitor websites. NOT provided by the data subject.
Retention: Indefinite today. A retention limit is a policy decision, not a technical one — tools/purge_snapshots.py can enforce one once a period is chosen.
For counsel: LIKELY legitimate interests, which requires a balancing test, a transparency notice under Art. 14 (data obtained from a third party), and a route to object. NONE OF THESE EXIST YET.
Machine-written commentary about named third parties
Examples: One-line implications generated by a language model about named companies and, incidentally, named individuals.
Source: Generated by this system.
Retention: Indefinite today.
For counsel: The highest-risk category here. It is model-authored opinion about identifiable people and companies, retained indefinitely. Labelled in the UI and in exports as AI-graded, but that is a product control, not a legal one.
Captured competitor page content
Examples: Extracted text of competitor homepages, stored per capture.
Source: Public web pages.
Retention: Configurable via tools/purge_snapshots.py; unchanged captures no longer re-store the body.
For counsel: Copyright and database-right questions here are separate from data protection and also need review.
Terms version identifier: unpublished-draft ·
no contact configured