Answer Library
Your reusable answers, scored for freshness and confidence, with a coverage check for any RFP or security questionnaire.
Included with secronjob . A seat without it sees an upsell here instead of the page.
Overview
Answer Library is the self-updating corpus behind your RFP and security-questionnaire responses. Every answer is scored on two axes: freshness, how current it is, and confidence, how much to trust it. You build the corpus: entries are added on this page, and the answer-library-curator and the questionnaire drafter both read from it, so they share one honest source of truth instead of a stale document nobody owns.
What it is for, on the page. Under the headline, this page says what it produces, what makes it run, and where the output lands. Every Work page reads that line from one registry, so the three answers cannot drift from the page.
The curator does not populate the corpus. It reviews what is there on its schedule, folds in newly approved answers, dedupes and flags stale entries. It writes no new entries, and this page said otherwise until 2026-08-30 (H165). If nothing has been added here, nothing appears, however long you wait. The same is true of the questionnaire drafter and the warm cache, which both read this corpus.
How freshness and confidence work
The scoring is deterministic: the same corpus always scores the same way. Freshness decays on a half-life from the day an answer was last verified, so an answer nobody has re-confirmed in a year sinks on its own. Re-verifying it, one click, resets the clock and refreshes it. Confidence is separate: it rises with approval, how often the answer has been used, and where it came from (a verified doc outranks a draft). An answer that was never verified reads as unverified, not as a fabricated high score.
The coverage check
Paste an incoming RFP or security questionnaire, one question per line, and the library matches each against your corpus by term overlap. Each question comes back as answered (a fresh, confident match exists), needs-refresh (a relevant answer exists but is stale and should be re-verified first), or a gap (nothing in the corpus is relevant). This turns a 200-question questionnaire into a short worklist: refresh a handful, write a handful, and the drafter handles the rest.
Nothing is invented
A question with no relevant, current answer is a gap, shown as a gap. The library never fabricates an answer to hit a coverage number, because a wrong answer in a security questionnaire is worse than an honest "we need to write this one."