Security questionnaires, answered once
Your next customer sent a 300-row vendor security assessment. You've answered most of it before — in last quarter's SIG Lite, a CAIQ, and two custom spreadsheets. RFPKit turns those into one approved library and fills the new questionnaire from it, in the customer's own file.
Create a free workspace Read the method
Built for the questionnaires you actually get
- SIG Lite and SIG Core spreadsheets from Shared Assessments licensees. RFPKit doesn't ship SIG content; it works with the file your customer sends. See how to answer a SIG Lite.
- CAIQ and CAIQ Lite from the Cloud Security Alliance, organized by Cloud Controls Matrix domains. Our CAIQ Lite walkthrough has a starter checklist.
- Custom vendor assessments — the spreadsheet a customer's security team wrote themselves, with merged section headers, drop-downs and a comments column. Map the question and answer columns once per sheet and you're done.
How a questionnaire gets answered
- Upload the customer's .xlsx. Pick the header row, the question column and the answer column for each sheet. All-caps and bold-only rows are proposed as section headings; one click turns them back into questions.
- Review the matches. Each question shows its three closest approved answers with a score and a High / Medium / Low label. Matching combines meaning (embeddings) with keywords, so “Is stored data encrypted?” finds “Do you encrypt data at rest?”.
- Use, draft or assign. Use a match as-is, ask for an AI draft built only from approved answers, write your own, mark it N/A, or assign it to the engineer or lawyer who knows. Keyboard shortcuts (J/K, 1–3, D, A) keep you moving.
- Review and export. Reviewers approve or send answers back with a comment. Export writes into the original workbook — only the answer cells change — plus a CSV and a JSON record of which approved answer and version produced each response.
AI first drafts that show their sources
Security answers have to be true. RFPKit's drafts are written only from the approved answers that matched the question (up to three, and only if the match is strong enough). The model must cite a source for every sentence; sentences without one are rejected and the draft is retried once, then flagged. If nothing in your library answers the question, there's no draft at all — the question is marked as a gap so you can assign it. Every draft carries the label “AI draft from your approved answers — check before sending.”
Drafts are capped at 300 per workspace each month to keep the service free; matching is not capped.
Keep answers current
Each approved answer has an owner and a review interval of 90, 180 or 365 days. When an answer is due, it shows up in the owner's review queue and in an optional calendar feed (ICS) they can subscribe to — no email. “Still accurate” resets the clock; editing creates a new version you can compare with the old one.
Honest limits
RFPKit doesn't certify anything. It won't give you a SOC 2 badge or a security score, and it won't submit answers into a customer's portal. It helps you answer consistently and quickly from what your team has already approved — you still own every answer you send.
Answer your next questionnaire from your library
Free, unlimited users and questionnaires. No account or email needed — you get a private link.
Create a free workspace