Build a data room with AI — describe the folder structure, AI builds it, you share it NDA-gated
Start with the shape of the room, not the pile. Describe the folders you want — the standard 8-category due-diligence tree (Corporate, Financial, Legal, IP, Team, Product, Cap Table, Tax), or something specific like "an executive summary folder, then one for interview booklets" — and the AI generates it for you to review before it lands. Then upload into it: a whole Google Drive or desktop folder goes up in one go with its nesting preserved. AI room generation can build a room straight from a document dump; AI document Q&A lets reviewers ask questions of the documents. You set permissions and share NDA-gated links. Median setup from account creation to first upload is 4 minutes 19 seconds. $52 per admin per month on Peony Data Room — unlimited documents, rooms, storage, and viewers.
See the output for yourself: browse a live example room built on the Seed Round template — the organized structure, exactly as investors see it.
Three ways to build the folder structure
By hand, by uploading a folder tree you already have, or by describing the structure you want to AI
Create the folders yourself, one at a time
Add content → New folder, one folder at a time. It gives you exact control over naming and over the order things appear in.
It is also the slowest route to a finished room — the other two methods exist so you rarely need this one.
Upload a folder tree you already have
Add content → Folder upload takes an entire folder from your desktop or a Drive export, nesting included, instead of making you rebuild the hierarchy by hand.
This is the one to reach for when the deal files are already organized somewhere else.
Describe the structure and let AI build it
An empty room offers Use template next to Add items, and that is the entry point. It opens Build with AI rather than dropping a fixed skeleton on you, so the structure you end up with is the one you actually asked for.
Describe it in plain language — "01. Corporate, 02. Financial, 03. Legal…", or "an executive summary folder, then one for interview booklets", or just "healthcare due diligence" — and it generates that tree. You review the result and choose to add it or retry, so nothing lands in the room until you approve it.
What does it mean to build a data room with AI?
Building a data room with AI means you skip the manual folder assembly. Instead of creating an eight-folder tree by hand, naming each folder and getting the numbering right — the part founders and advisors dread — you describe the structure you want and the AI generates it. That can be the standard 8-category due-diligence structure — Corporate, Financial, Legal, IP, Team, Product, Cap Table, and Tax — or anything the deal actually calls for: an executive summary folder followed by interview booklets, a builder due-diligence tree, a healthcare due-diligence tree. You review the generated tree, add it, and upload into it.
On Peony there are three distinct AI capabilities, and it's worth keeping them straight. AI folder-structure generation turns a written description into the folder tree, which is the one this page is about. AI room generation stands up a room end-to-end from a document dump when you want the room built rather than folder-by-folder. AI document Q&A lets a reviewer ask a question and get an answer drawn from the documents in the room — grounded in what's been shared, not the open internet, so a buy-side reviewer can trace every answer back to a source they can open. AI folder-structure generation and AI room generation are on the Data Room plan at $52 per admin per month; AI document Q&A is on the Business plan at $30 per admin per month.
The honest boundary is the same one that makes the AI trustworthy: it categorizes, and you verify. On a typical upload the AI sorts the clear majority correctly and misses a small handful of edge cases — a document that belongs in two categories, a cryptically named file, a one-off the standard structure doesn't anticipate. Moving those strays takes about a minute; the final structure is your responsibility, not the model's. What the AI removes is the hours of manual sorting, not the five-minute review — and it should not remove that review. This is why a chat tool alone isn't a data room: a chat tool can suggest an outline, but only the room itself can host the files, gate them behind an NDA, watermark each page, and prove who read what.
What does the AI actually do?
AI folder-structure generation and AI room generation run on the Data Room plan at $52/admin/month; AI document Q&A runs on the Business plan at $30/admin/month. Across every AI step the pattern is the same: the AI drafts, you review and confirm before you share. The final structure is yours.
Who builds a data room with AI?
How to build a data room from a Drive mess
- 1
Gather, don't organize
Pull every file into one place — cap table, financials, contracts, IP, deck. Do not build folders first; that's the work the AI removes.
- 2
Bulk-upload the whole pile
Drag and drop a Google Drive export or a desktop folder in one go. An exported Drive or SharePoint hierarchy is preserved on upload.
- 3
Describe the structure
Say what folders the deal needs and the AI generates the tree — standard 8-category, or whatever your process actually uses.
- 4
Review, then fill it
Check the generated tree before you add it, then upload your documents into the folders. The AI drafts; you confirm the final room before sharing.
- 5
Gate and share
Turn on dynamic watermarks and the Advanced NDA with countersigning, set granular permissions per investor or bidder group, and send NDA-gated links.
When is AI not the part that matters?
When you only need a naming outline. If all you want is a suggested folder list, a general chat tool can draft one. Organizing is the easy 10% — but a chat tool can't host, gate, watermark, or log access, so you'll still need a room to put the files in. The AI is the setup speed; the room is the product.
When the room is tiny and low-sensitivity. A single document sent to one reviewer once doesn't need a generated folder tree at all — a watermarked personalized link is enough. Don't build a full room for a one-file share.
When you need Q&A but not watermarks yet. If your priority is letting reviewers self-serve answers and the room is low-sensitivity, AI document Q&A on the Business plan at $30/admin/month is a legitimate cheaper start — structure generation and watermarking come with Data Room at $52.
When the structure has to be exact and non-standard. The AI produces the standard 8-category tree. If your firm's playbook requires a specific non-standard structure, you'll build that yourself on top of the AI's baseline — the AI gets you to the standard fast, your template layers over it.
When you'd skip the review. The AI categorizes, you verify — that's not optional. If you're not willing to open every folder and move the few strays before sharing with investors, no tool will save you. The final structure is your responsibility.
"Peony has been great for sharing documents with investors, employees, and customers. It's easy to use, good value, and new features are constantly being added. Definitely recommend!"
Ed Harris
Founder & CEO, Ligo Bio (YC S24)
Frequently asked questions
We're a 4-person startup with a term sheet meeting Monday and our docs are scattered across three founders' Google Drives and someone's desktop. Can AI actually build us an investor-ready room by then?
Yes, and the weekend is more than enough. Start with the structure rather than the files: describe what you need — "01. Corporate, 02. Financial, 03. Legal, 04. IP, 05. Team, 06. Product, 07. Cap Table, 08. Tax" — and the AI generates that folder tree for you to review before it lands in the room. Then pull everything into one place (cap table, SAFEs, financial model, incorporation docs, IP assignments, pitch deck) and upload it into the folders. Median setup from account creation to first upload is 4 minutes 19 seconds. The AI that builds the structure is on the Data Room plan at $52/admin/month, which is also where dynamic watermarking, Advanced NDA with countersigning, and granular permissions live — so by Monday you have a numbered, NDA-gated room instead of a Drive link that reads as unprepared. The honest part: the AI drafts the shape of the room, you decide what goes in each folder. Budget the weekend evening for filing and renaming — the final structure is your call, not the model's.
I'm a solo M&A advisor running 3 concurrent sell-side listings. I don't have time to hand-build a folder tree three times. How does the AI cut that down?
Each listing is its own room, and each one starts the same way: describe the structure once, let the AI generate the folder tree, then fill it. Instead of assembling three diligence indexes by hand you reuse one description across three rooms and get the same numbered tree every time. On the Data Room plan at $52/admin/month you get unlimited rooms, so three concurrent listings cost the same as one, with dynamic watermarks stamping every page with the bidder's identity and granular permissions segregating what each bidder track sees. AI room generation can stand up a room straight from a document dump when you want the room built end-to-end rather than folder-by-folder. Where a solo advisor still carries the load: the AI does not know your deal's confidentiality tiering — which bidder sees the customer contracts versus only the teaser — so you set the visitor groups. The AI gives you back the hours you'd have spent dragging files into folders; the judgment stays yours.
I'm a CFO with a single 140-file Google Drive folder we've used for two years, and we're opening a Series A. It's a mess of naming conventions. What does the AI do with something that disorganized?
Two things help, and neither of them is magic file-sorting. First the structure: describe the tree you want and the AI generates it, so you are not naming eight folders by hand or inventing the taxonomy from scratch. Second the move: download the Drive folder — the hierarchy is preserved on export — and Folder upload takes the whole tree in one go, nesting included, so whatever organization does exist survives. From there you drag the strays into the right folders. One watch-out on Drive exports: Google can rename duplicate files, so spot-check names after upload. This runs on the Data Room plan at $52/admin/month, which adds dynamic watermarking, Advanced NDA with countersigning, and granular permissions for staged investor access. Honest limit: the AI builds the shape of the room; it does not read your 140 files and decide where each one belongs. That placement is yours. Expect a focused half-hour on a folder that disorganized — still far less than a day spent inventing the structure and filing at the same time.
Skeptical question: can't I just paste my files into ChatGPT and have it organize them? Why pay for a data room?
For naming ideas and a proposed folder outline, yes — a general chat tool can suggest a structure, and there's no shame in using it that way. But organizing is the easy 10% of the job. A chat tool cannot host the files behind a link, gate them behind an NDA, watermark each page with the viewer's identity, block and log screenshots, or tell you which investor read the cap table for eight minutes and skipped the roadmap. The room is the product; the AI is the setup speed. On Peony the AI drafts that same structure inside the place the documents actually live and get shared — you describe the tree, it generates it in the room — and then Data Room at $52/admin/month wraps it in dynamic watermarking, Advanced NDA with countersigning, granular permissions, and page-level analytics. So the honest framing is: use a chat tool to brainstorm the outline if you like, but you still need somewhere to put the files that can enforce who sees what and prove who saw it. That somewhere is the data room.
I'm a PE associate and I'm trying to standardize the diligence rooms across our portfolio companies so they all look the same. Does AI room generation help me enforce one structure?
It does, because the structure comes from a description you control, so you can run the identical one across every portfolio company regardless of how each handed you their files. One company's tidy SharePoint and another's chaotic desktop both end up behind the same numbered tree, which is what you want when a deal partner opens six portfolio-company rooms and expects them to look identical. AI room generation can stand a room up from a document dump when you're onboarding a new company from scratch. This runs on the Data Room plan at $52/admin/month with unlimited rooms, so standardizing across a portfolio doesn't multiply your bill, and granular permissions plus dynamic watermarking apply uniformly across every room. Where you still steer: the AI produces the structure you asked for, but any portfolio-specific folder your firm's playbook requires — a particular ESG or KPI tab — you write into the description or add yourself. The AI gets every room to the same baseline fast; your template layers on top.
First-time founder here — what does the AI actually get wrong when it organizes my files? I don't want to blindly trust it in front of investors.
Good instinct, and the honest answer is that the AI categorizes and you verify — the final structure is your responsibility, not the model's. In practice it sorts the clear majority of a startup document set correctly into the standard 8 categories in under 3 minutes, and the misses are usually edge cases: a document that legitimately belongs in two categories (an advisor agreement that's both Legal and Team), a file with a cryptic name it has to guess from, or a one-off your company has that the standard structure doesn't anticipate. On a typical upload that's a small handful of files to move, which takes about a minute, plus a few minutes to rename anything unclear. It also flags categories that look thin, which is a feature, not an error — that's how you catch a genuinely missing document versus one the AI merely filed elsewhere. So don't blindly trust it: open every folder, move the few strays, and rename for clarity before you share. The AI removes the hours of manual sorting; it doesn't remove the five-minute review, and you shouldn't want it to.
I'm an advisor and I want reviewers to be able to ask questions of the documents without emailing me. Which plan has the AI Q&A, and is that the same thing as the AI that builds the folder structure?
They're two different AI capabilities on two different plans, so it's worth being precise. AI document Q&A — where a reviewer asks a question and gets an answer drawn from the documents in the room — is on the Business plan at $30/admin/month ($44 monthly). The AI that builds your folder structure from a plain-language description is on the Data Room plan at $52/admin/month ($75 monthly), alongside dynamic watermarking, Advanced NDA with countersigning, granular permissions, custom domain, and AI room generation. So if your priority is letting reviewers self-serve answers, Business covers it; if you also want the AI to build the folder structure and you want watermarking and the Advanced NDA, you're on Data Room. Most advisors running real diligence end up on Data Room because the watermarking and countersigning NDA matter as much as the Q&A — but if the room is small and low-sensitivity, Business with AI document Q&A is a legitimate cheaper starting point.
Buyer-side question: if the reviewers use AI Q&A, does it answer from the documents in the room only, or is it pulling from the open internet? I need answers scoped to what's actually been shared.
Scoped to the room. AI document Q&A answers from the documents in that data room — a reviewer asks a question and the answer is drawn from the files that have been shared, not from the open web, so it can't invent facts the seller never put in front of you. That matters on the buy-side for exactly the reason you're asking: an answer is only useful in diligence if it traces back to something in the room you can then open and verify yourself. AI document Q&A is on the Business plan at $30/admin/month; the Data Room plan at $52/admin/month adds dynamic watermarking, Advanced NDA with countersigning, granular permissions, AI folder-structure generation, and AI room generation on top. The honest caveat: grounded-to-the-documents means the AI is only as complete as what's been uploaded — if a document isn't in the room, the AI won't answer from it, which is a feature for trust but a reminder to confirm the index is complete before you lean on the Q&A. Treat it as a fast way to locate and summarize what's there, then read the source document for anything that moves the deal.
Related resources
AI-powered data rooms
How the AI turns a written description of the folders you want into the structure of the room.
Data rooms
The room itself: NDA gates, dynamic watermarks, granular permissions, and per-viewer audit trails.
Startup data room checklist
The 60-document, 8-category standard VCs expect — the structure the AI builds toward.
Seed round data room
The tab-by-tab build spec for a seed raise — the operator core the AI-built room starts from.
Pricing
Free, Business at $30, and Data Room at $52 per admin per month — which AI capability lives on which plan.
All use cases
Data rooms by file type and workflow — CAD, BIM, DICOM, drone, point cloud, geological, and more.