Guide
How to make a data room AI-ready for due diligence
Organize the data room with a clear index, text-based files, consistent names and an explicit version for every document, and make the key figures, dates and parties readable as text with their qualifiers. Buyers will review it with AI, so give each document a fidelity-checked, AI-readable version with links back to the source page, slide or cell — that way their tools cite your numbers rather than guess them.
Buyers already review with AI
On the buy side, AI is used to triage hundreds of documents: find change-of-control clauses, pull revenue by customer, compare the model with the management accounts, and flag inconsistencies. Each of those steps re-extracts content from your files.
If the extraction is wrong, you pay for it in questions, delays and price discussions. A data room that reads cleanly to machines shortens the Q&A rather than adding to it.
Prepare the room
- Use a numbered folder structure that mirrors a standard request list: corporate, financial, commercial, legal, HR, IP, tax.
- Name files with what, status and date: “Customer contract — Acme — executed — 2025-06-01”.
- Prefer text-based exports. Replace scans with text-based copies where you can; mark scans that remain.
- Upload spreadsheets as spreadsheets, not as PDF printouts, so formulas and inputs stay inspectable.
- Remove drafts and duplicates, or label them clearly as superseded.
- Keep a written index of every document with a one-line description.
Make the numbers consistent across documents
AI review is good at finding disagreements between documents. Find them first.
| Check | Why it matters |
|---|---|
| Revenue in the model vs management accounts vs CIM | Different figures look like errors even when the basis differs |
| Metric definitions (ARR, GMV, net revenue) | State each definition once and use it everywhere |
| Periods and fiscal year ends | Label every column with dates, not only “FY” |
| Adjusted and pro forma figures | Keep the adjustment next to the number |
| Contract values vs customer list | Amounts, currencies and dates should match the signed contracts |
Versions and Q&A
- Each document should have one current version. When you update, replace rather than add “v3 final final”.
- Record what changed and when, so the buyer can see the history.
- Link Q&A answers to the documents and pages they rely on.
- Tell the buyer which documents are the authoritative source for key figures.
Where an AI Data Room fits
An AI Data Room in The Papers is a permanent address that holds a company’s documents, each with a canonical AI Version — a page for people plus Markdown and JSON, and a room index (llms.txt and index.md) so AI tools can find every document.
Before a document is published, every number, currency, date and period must match the source; spreadsheet formulas keep their dependencies; qualifiers stay attached; and each value references its source page, slide or cell. Ambiguities go to you to confirm. Each document gets an AI Readiness score and a status, and documents can be private to Data Room members, with Team Members per room.
Updating a document keeps its link and history, and analytics show which people and AI agents read each document. It is a hosted service and does not certify that the content is true — only that the AI Version is faithful to your source.
FAQ
- Can I use an AI Data Room alongside our existing VDR?
- Yes. You can keep your existing room for permissions and Q&A workflows and share fidelity-checked AI Versions of key documents next to it.
- Will buyers’ AI tools respect access controls?
- Private documents are only available to Data Room members, and unlisted documents are never indexed. What a buyer does with content they can open is governed by your NDA.
- Should we upload the financial model?
- Yes, as the original spreadsheet. The Papers preserves cell values, number formats, formulas and their dependencies, and named ranges.
- What if two documents disagree?
- Fix it at the source before publishing, or explain the difference in basis in both documents. Fidelity check checks each document against its own source, not against other documents.