This page is written for the security or compliance reviewer who has to approve a new vendor before anyone uploads a document. If a question is not answered here, email hello@rfparse.com and we will answer it in writing.
The RFP documents you upload, kept so that every risk and compliance flag can be traced back to the exact page and clause it came from.
The analysis generated from each document, as structured data: scores, requirements, risks, strategy, and the evidence quotes behind them.
Your company profile (capabilities, certifications, markets) and pipeline outcomes (won, lost, passed), which are used to personalize and calibrate scoring.
Account identity (name, email) via Clerk, and subscription status via Stripe. RFParse never sees or stores card numbers.
Analyses and profiles are stored in a Postgres database hosted by Supabase. Uploaded documents are stored in a private Supabase Storage bucket under a path scoped to your account.
Documents are never publicly addressable. They are served only through short-lived signed URLs (five minutes) issued to the signed-in owner.
Every database query and file access is scoped to the requesting user. Read-only share links you create are signed, expire after seven days, and expose the analysis but not the source document or exports.
Anthropic: document text is sent to the Claude API to produce the analysis. Under Anthropic’s commercial API terms, inputs and outputs are not used to train their models.
Supabase: database and file storage. Clerk: authentication. Vercel: application hosting. Stripe: billing.
RFParse does not use your documents or analyses to train models, does not sell data, and does not share it with anyone outside this list.
Documents and analyses are retained until you delete them. Deleting an analysis in the app permanently removes both the analysis record and the stored document.
To remove your account and all associated data, email hello@rfparse.com and we will confirm when deletion is complete.
RFParse is a screening aid, not a compliance opinion. Every analysis is labeled as AI-assisted output requiring human review.
Each analysis is stamped with the model that produced it, the time it ran, and the inputs used, so a decision can be reconstructed later.
Evidence quotes are verbatim from your document. If a claim has no quote, treat it as the model’s inference rather than something the RFP states.
See also the Privacy Policy and Terms of Service. Last updated September 2026.