Last verified 2026-07-12

DDQ & RFP answering from an approved answer bank

Can AI answer investor DDQs and RFPs? Yes for the draft — matching each question to your approved answer bank and writing a first response in your house language. No for the send: an outdated or subtly-wrong answer to an investor is a representation, so a human still approves every one.

The pain	During a raise a manager fields a stream of due-diligence questionnaires and RFPs — fundraising, operational, ESG — each asking the same 200 things in a different order and format, some as portals, some as spreadsheets, some as Word. An IR or compliance person hunts the last approved answer, pastes it, adjusts it to the question, and repeats, hundreds of times per document, while the flagship raise waits on the turnaround.
What AI does today	Match each incoming question to your bank of previously-approved answers and draft a response in your own wording, flagging where no approved answer exists and where two answers conflict. The boundary: the semantic match ("describe your BCP" = "outline business-continuity arrangements") is the AI slice; the answer bank itself, the approval workflow and the version control are ordinary content management you may already run.
Proof it's real	Vendor-claimed, established category. Purpose-built DDQ platforms for private-markets IR are an established product category — Ontra DDQ, for one, is built around an approved-answer library for managers fielding 15–30+ questionnaires a week in a flagship raise. Client counts and time-saved figures are the vendors' own; the category's existence and adoption in IR teams is not in doubt. Newest evidence: 2026.
What it can't do	It cannot approve the answer — that is where the human sits: a DDQ response is a representation to an investor and often to their regulator, so a person confirms each material answer is current and correct before it leaves, and every one traces to an approved source. An answer that was true last fund and is now stale reads as accurate and isn't.
The real alternatives	The IR team and the shared drive — the status quo: past DDQs in a folder, searched by hand. Free, and workable until volume spikes; the cost is turnaround time mid-raise. A DDQ/RFP platform (Ontra/Dasseti/DiligenceVault-class) — the answer bank, workflow and AI drafting as a product; wins for managers with standing questionnaire volume. An answer bank on the tools you already run — SharePoint plus an internal LLM gateway does the matching-and-drafting job for teams whose group restricts external vendors; the curation burden is the same either way. IR consultants for the raise — surge capacity, priced per engagement; wins for a one-off flagship raise without standing volume. Whichever route: the curated answer bank is the asset — the AI is only as good as the library it drafts from.
What you need in place	A curated, dated, owned answer bank (building it is the real project — the tool is secondary); a compliance approval workflow with staleness review dates on every answer; and ownership — IR runs the process, compliance owns what goes out. If your answers live in twelve people's sent folders, start there, not with a vendor. And in a group, client-facing text usually needs central-compliance sign-off wherever it's drafted — build that approval path in from day one.
Effort & cost	Weekend script — an LLM with your past approved DDQs pasted in, drafting answers to a new questionnaire for you to check; free, immediate on a live raise. Off-the-shelf — DDQ platforms, typically five-figure annual subscriptions. Real project — an answer bank wired to your document store with approval workflow and staleness flags; five figures plus the curation effort, which dominates. Fielding a handful of questionnaires a year? The shared drive plus a good index is honestly fine.
What to watch	Staleness is the risk, not blank answers — the tool will confidently reuse an answer that a policy change quietly invalidated. Keep the answer bank dated and owned, and watch for answers pulled from a source past its review date.

Questions operators ask

Can AI fill in a DDQ by itself?

It can draft the whole questionnaire from your approved answer bank — that part is real and shipping. What it cannot do is warrant the answers: each material response is a representation, so human approval per answer is the control, and the audit trail of who approved what is your examiner answer.

Do we need a DDQ platform, or is our shared drive enough?

Volume decides. A handful of questionnaires a year: the shared drive with a maintained index is fine. Fielding them weekly mid-raise: the platform's answer bank, workflow and semantic matching pay for themselves in turnaround time — but only after you've curated the library it drafts from.

Changelog

2026-07-12 — published.