Turn a folder of contracts into a queryable spreadsheet
Review vendor portfolios, contract repaper, and diligence sets at spreadsheet speed. Upload up to 200 documents, define the questions you need answered as columns, and every cell comes back grounded in a source quote with page citations.
What You Get
Spreadsheet-style multi-document review
Rows are documents, columns are the questions you need answered. Sort, filter, comment on cells, and export, with AI doing the reading.
Source-quote grounded cells
Every cell links to the exact source passage with page and paragraph citations. Click any cell to see the supporting quote in context.
Prebuilt column templates
Start from NDA, SaaS MSA, commercial lease, employment, or IP license templates and customize columns per matter.
Bulk upload with OCR
Drag in up to 200 files (PDF, DOCX, TXT, XLSX). Scanned PDFs auto-route through OCR. Files appear as rows as they complete.
Free-text and structured columns
Ask custom questions as free text, or use structured types (date, number, currency, multi-select). Each cell answers within its document, not across the corpus.
Export and share with co-counsel
Export to CSV, XLSX, or PDF. Share matrices via the same folder access used for chats and drafts.
How Document Matrix Works in Vaquill AI
Why this matters
Mechanical extraction consumes 60-80% of associate time on document-heavy matters. A 500-contract diligence can run 6-8 weeks today.
- ✓M&A diligence: extract change-of-control, termination, IP assignment, and liability caps across hundreds of contracts.
- ✓Contract portfolio: find every agreement that expires in Q3 2026 or carries MFN clauses in seconds.
- ✓Lease abstraction: 20-50 data points across 200 leases, two days instead of two weeks.
Hallucination prevention
Every claim is anchored to text in the source document.
- ✓Quote-grounded prompts: the LLM must return the supporting passage alongside the extracted value.
- ✓Optional claim verifier checks that the answer is actually supported by the cited quote.
- ✓Cells where the document does not address the column return "Not found", never a guess.
How it fits with the rest of Vaquill AI
Matrices share folders, sharing, and retrieval with Chat, Drafts, and Research.
- ✓Matrices live in the same folder tree as chats and drafts. Organize by matter, client, or practice group.
- ✓Org-level sharing: any teammate with folder access can view or co-edit the matrix.
- ✓Credit-based pricing with pre-run cost estimates so a 200-document run never runs away from you.
How It Works
Get started in minutes, not weeks.
Create a matrix
New Matrix in the sidebar, name it, save under a folder. The empty grid is created and added to the folder shared with your chats and drafts.
Bulk upload documents
Drag in up to 200 PDFs, DOCXs, TXTs, or XLSXs. Scanned PDFs auto-route through OCR. Files appear as rows as they complete.
Define columns
Pick a prebuilt template (NDA, MSA, lease, employment, IP) or add custom free-text columns. Each column is one question you want answered from every document.
Run extraction
Vaquill AI runs extraction across all rows and columns in parallel. Cells populate live with source quotes. Pre-run cost estimate before you commit.
Review, export, share
Filter, sort, comment on cells, correct any errors inline. Export to CSV, XLSX, or PDF. Share the matrix with co-counsel via folder access.
Frequently Asked Questions
How is Document Matrix different from Hebbia Matrix or Kira?
Hebbia Matrix is enterprise-only at $40K+/yr. Kira requires dedicated onboarding and is priced for AmLaw 200 firms. Vaquill AI offers the same grid pattern, self-serve, at solo, small-firm, and in-house pricing, with credit-based scaling so a 50-document run is affordable.
How many documents can I put in a matrix?
Up to 200 documents per matrix in v1. Larger sets are split across matrices today; multi-matrix dashboarding is on the roadmap.
What document formats are supported?
PDF, DOCX, TXT, and XLSX. Scanned PDFs are automatically routed through OCR before ingestion, so image-only contracts work too.
How do I trust the cell answers?
Every cell links to the exact source passage with page and paragraph citation. Prompts are quote-grounded, so the LLM must return supporting text alongside the extracted value. Cells without source support return "Not found" rather than a guess.
Does the matrix update if I add new documents?
Yes. Upload more documents to an existing matrix and only the new rows run extraction. Existing rows are unchanged.
Can I share a matrix with my team?
Yes. Matrices live in the same folder tree as chats and drafts and inherit the same role-based access. Teammates with folder access can view or co-edit.
How much does it cost to run a large matrix?
Cells are billed at 1-3 credits each depending on extraction complexity. You see a pre-run cost estimate before you commit, and per-matrix credit caps prevent runaway spend.
Are there prebuilt templates I can start from?
Yes. NDA, SaaS MSA, commercial lease, employment agreement, and IP license templates ship with v1. You can customize the columns per matter or save your own column sets as reusable templates.
Stop copying clause values into spreadsheets
Upload your document set, define your columns once, and get a sortable, citation-grounded matrix in minutes instead of weeks.