ChatDocument

AI document analysis that shows its sources on every page

Analysis means more than one answer. You need to compare versions, pull the same fields from a stack of files, spot what is missing and hand a result to someone who will check it. ChatDocument brings these jobs into one workspace, and every result, from a single sentence to a table cell, links to the page it came from.

Try it on your PDF
Three colleagues at a meeting table comparing printed reports, one pointing at a highlighted paragraph
p. 14

Try it on your PDF before analysing a whole folder

The chat with PDF demo runs one question on one PDF or on a ready document such as the quarterly report. It shows the core of every analysis in ChatDocument: an answer written from the file, with the page highlighted. No card needed for the demo.

Four ways to analyse documents with AI

Each tool works on the documents in one space and cites its sources the same way.

  • Ask: plain questions across every file in the space, with document name and page per statement
  • Compare: two to five documents side by side, for example two contract versions or three supplier offers, with the differences cited on both sides
  • Extract to table: custom columns such as party, start date, fee and notice period, one row per document, every cell linked to its page, export to XLSX or CSV
  • Summarize: one page, section by section or with your own template, each line cited

AI document intelligence that holds up in review

An analysis is only as good as the review it survives. ChatDocument checks every cited passage in code: it must exist on the cited page and contain the quoted text before the result is shown. A reviewer clicks a cell or a sentence and lands on the highlighted passage. On Business and Enterprise, the audit log records who asked what and which pages answered, so the trail stays with the work.

  • Verified citations on answers, comparisons, tables and summaries
  • Slide numbers for decks and sheet with cell range for spreadsheets
  • Audit log on Business and Enterprise

Why generic AI for documents falls short in analysis

A general chatbot with an upload button blends your files with what it already knows, and the blend is invisible. In analysis that is a real risk: a missing indemnity cap can be reported as present because most contracts have one. ChatDocument answers only from your selected space. When a document lacks something, the result says it is not there and lists the closest passages checked, so an absence is a finding, not a guess.

Where teams run document analysis

Plans for AI document analysis

Yearly billing is the default and saves 50%. One row of an extraction (one document) counts as 1 question, a comparison counts as 3.

PlanYearly billingMonthly billingAnalysis limits
Starter 24 USD per month (288 USD per year) 49 USD per month 500 questions, compare, extract 10 documents per run
Team 74 USD per month (888 USD per year) 149 USD per month 2,000 pooled questions, extract 50 documents per run, roles
Business 249 USD per month (2,988 USD per year) 499 USD per month 7,000 questions, extract 250 per run, audit log, API
Enterprise 745 USD per month (8,940 USD per year) 1,490 USD per month 20,000 questions, extract 1,000 per run, SSO (SAML), SLA 99.9

Details on chat with PDF pricing.

FAQ

Which file types can be analysed

PDF (including scans read with OCR), Word (DOCX), Excel (XLSX, CSV), PowerPoint (PPTX), plain text, Markdown and web pages by URL.

Can the analysis use files from our shared drive

Yes. Connect Google Drive, SharePoint and OneDrive or Dropbox, select the folders, and they sync into the space on a schedule.

What happens to documents after the analysis

They stay in the space until you delete them. Deleting a document or the space removes the document and its index, and the data leaves backups within 30 days.

Does deleting documents give questions or pages back

No. Pages count once per billing period and questions count when they are answered, so deletion does not return usage.

Start the analysis with one file

See a cited answer on your own PDF, then bring the rest of the folder.

Try it on your PDF