AI Auto-Indexing: How It Works
Peony's AI classifies and organizes an unstructured batch of files into a logical data room folder structure in under three minutes.
Last updated April 10, 2026
Drop a messy folder of files into a new Peony data room and the AI reads each document, classifies it, and places it into a logical folder structure. Standard rooms finish in under three minutes.
What It Does
Auto-indexing runs the moment uploads finish. It identifies each document's type (financial, contract, deck, cap table, and so on) and moves it into a matching folder, creating new folders when needed.
It's on by default for every new data room and runs in the background, so it doesn't block link setup, watermarks, or NDA gates.
How to Use It
New rooms: nothing to turn on. Upload files and the indexer runs.
Existing rooms: after adding new files, open the AI menu and click Re-index. New files get classified into the existing structure; new top-level folders are added if new document types appear.
To take full manual control, disable auto-indexing in data room settings. Most teams leave it on and just drag files if they disagree with a placement — the AI will not move them back.
What's Classified
- Financials — P&L, balance sheets, cash flow, quarterly/annual reports, budget models
- Legal — contracts, NDAs, IP filings, employment agreements, formation docs, board resolutions
- Customer and revenue — pipeline, customer contracts, CRM exports, cohort analysis, churn
- Product — specs, roadmaps, technical docs, architecture diagrams
- Team — org charts, resumes, offer letters, benefits
- Market — TAM analysis, competitor notes, industry reports
- Fundraising — decks, cap tables, term sheets, SAFEs, convertible notes
When a document fits multiple categories, the indexer picks the dominant one. Drag to override.
When It Might Miss
- Non-English documents. Coverage is limited today. Expect manual cleanup.
- Heavily custom file formats. Proprietary or unusual templates may not classify correctly.
- Very sparse folders. Two files with no context may land in "Other". Add more files or rename them descriptively.
- Scanned images without OCR. Peony runs OCR on most scans, but very low-quality scans can slip through.
Drag misclassified files to the correct folder — the indexer won't override manual placements on the next run.
Privacy Note
Classification runs on Peony's own infrastructure. File contents are not sent to third-party AI providers and are not used to train any model. Results are stored as metadata; underlying content stays encrypted in Peony's storage.
If a buyer's security team asks about AI data handling, auto-indexing is an internal classifier, not a pass-through to an external LLM. For how Peony handles AI Q&A (which does use a production LLM), see AI Q&A.