Trust
Where your data goes, what we do with it, and what we refuse to hold. Short, and only what the code backs.
What we store
Narrative Pro keeps four kinds of content: your account and profile (name, email, and the service branch and level you choose to give); the narratives you save, with their earlier versions; the résumés you build; and the drafting conversations you have with the assistant.
All of it lives in one Postgres database hosted by Supabase. Netlify serves the site. Stripe handles payment — we never see or store your card number. A short-lived server cache holds recently used records to keep pages fast; it is not a second copy of your data.
Content stays until you delete it or close your account.
Encryption in transit
Every request between your browser and narrative-pro.com travels over HTTPS. Our server adds a Strict-Transport-Security header (one year, subdomains included, preload) so a browser
that has visited once will refuse to talk to us over plain HTTP.
Encryption of stored data is provided by our hosting providers; we do not restate their figures here. See Supabase's and Netlify's own security documentation.
Your controls
Export. In Settings → Privacy, submit an access or portability request. It is fulfilled immediately: a JSON copy of your profile, narratives, résumés, conversations and usage records is attached to the request for you to download.
Delete. In Settings, Delete my account erases the account and the content listed above. It runs when you confirm it; it is not queued behind a review.
Stop AI processing. The AI processing switch in Settings → Privacy turns off every assistant feature. While it is off, nothing is sent to the model.
AI processing
Assistant features send the text you are working on to Google's Gemini API. Every call passes through one server-side gateway; the model SDK is imported nowhere else. The gateway reserves against your daily token cap before the call is made, so usage is bounded per account.
We never train models on your content. We have no training or fine-tuning pipeline. Your text is used to produce your output, and for nothing else.
Google's API data-use terms apply to the model call; see our privacy policy.
Classified and CUI material
Narrative Pro is for unclassified, unmarked material only. It is not authorised to hold classified information, Controlled Unclassified Information (CUI) or export-controlled material, and none of the services named above are approved for it.
We enforce that at the input. Text carrying a classification or CUI marking — a banner line, a portion mark such as (S) or (TS), CUI, FOUO, NOFORN, a classification block — is refused before it is saved and before it is sent to the model. The refused text is not stored and not logged. Text that only mentions a sensitive system, or a word like OPSEC, gets a reminder, not a refusal.
Keep operational detail out: readiness figures, deployment specifics, program names. Describe the outcome and your part in it. That is good OPSEC and, as it happens, better bullet-writing.
Incident contact
If you think your data has been exposed, or you have pasted something you should not have, write to support@narrative-pro.com or use the contact page. Tell us what happened and when. We follow the breach-notification terms in our privacy policy.
Last updated: 10 September 2026