Privacy Policy

• Your GEDCOM file — uploaded and processed in memory to run the scan. The scan runs on our server because analyzing every person, family and source in a full tree and producing your score is too heavy to run reliably inside a browser tab, especially for large files. For large files we briefly store it during the scan and request deletion as soon as the scan finishes. Deletion retries automatically after transient failures. It is used only to produce your results and the features you choose to run. If a large-file upload never reaches the scan step (for example, you close the tab first), the file is removed automatically by a scheduled cleanup, normally within about an hour.
• Tree details used by AI — only when you use an AI feature. Depending on the feature, this can include names, dates, places, spouse names, family labels, and issue evidence for relevant people. Roadmap sends evidence for up to 32 issues, and one issue can mention several people. See “Who we share it with” below.
• IP address and browser type — standard hosting and abuse-prevention data, plus optional Vercel page analytics on our informational pages only. You can exclude this browser from optional analytics at Analytics preferences.
• Optional product telemetry — when enabled for this browser, usage events are sent as you use the app: whether an upload was accepted or rejected — and if rejected, a general reason category (for example, wrong file type or file too large) — whether a scan succeeded, failed, or was canceled (plus how long it took, how many issues it found, whether it was the demo tree, and general categories describing anything unusual the scan noticed), which results tab you opened, which genealogy standard and severity an issue you opened relates to, whether an export succeeded or failed (in what type and format, how many issues it covered, and a general reason category on failure), and pricing-page views or button clicks. Most of these events carry no identifier at all. The one exception: once a real (non-demo) scan succeeds, opening an issue’s details or completing an export afterward in the same browser session sends a short code — not your name, email, or account — that lets us tell those two specific events came from the same session in the same calendar week, so we can measure whether people who complete a scan go on to explore their results. No other event carries this code — not the scan itself, and not simply opening a results tab. That code lives only in memory for the current page session, is never written to your browser’s storage, and cannot be linked across a page reload, a new session, or a different week. File size is recorded as a general range, not the exact size or filename. No names, family-history dates, places, or GEDCOM content are ever included. Each event does carry a timestamp of when it happened, used only to measure usage over time. See “How long we keep it” below.
• Place names — only when you view the Places map, sent to our geocoding provider to look up coordinates.
• Notify-me / interest signups — if you submit the “Notify me” form on our pricing page, we store your email address and your answer to whether you’re doing genealogy work for clients or working toward certification, in our database (Neon, US). Your Yes/Maybe/No answer about a proposed $39 report is counted only in weekly totals; it is not stored with your email. We use those totals to measure interest. If you instead submit the older product-updates waitlist form (being retired), we store your email address and optional name the same way. We use your email only to send the one announcement. Email privacy@kinsift.com to have it deleted.

Before the first AI request in a browser, KinSift shows a data-sharing notice and asks for consent. Roadmap removes details for likely-living relatives; Conflict Resolver may include names needed for the selected task. AI features are available on paid plans only, so Free users’ data is never sent to OpenAI.

• Your GEDCOM file: deletion is requested when the scan finishes, with automatic retries after transient failures.
• AI responses: cached on our servers for up to 30 days so repeat clicks don’t burn your daily quota.
• OpenAI API content: may be retained by OpenAI for up to 30 days for abuse monitoring.
• Optional product telemetry: when enabled, individual events pass through our hosting platform’s standard request logs and are folded into anonymous weekly counts that we keep for 13 weeks.
• Session cookie: 30 days (remembers your plan).
• Browser storage (scan history, theme preference): stays on your device until you clear it.

To remove browser-held KinSift data, clear your cookies and local storage. Use Analytics preferences to exclude this browser from optional analytics. The setting does not stop hosting and abuse-prevention logs, rate limiting, billing records, or server-verified interest and purchase totals. KinSift does not host share links or persist scan artifacts for sharing. Cached AI responses expire within 30 days. OpenAI abuse-monitoring records age out under OpenAI’s retention terms and cannot currently be deleted throughKinSift. A self-serve KinSift cache-delete option is planned.

Questions about your data or this policy? Email privacy@kinsift.com.