Documents and AI Insights
Upload the data room. Get back numbers with confidence scores and answers with citations.
Sixteen file types, scanned pages included. Vera detects the document type, extracts the financials, checks each year against the column it came from, and answers questions with numbered citations that open the exact passage.
- Every claim carries a numbered citation
- Opens the exact passage
- Answers come from the deal's own documents
Documents
- Sixteen file types: PDF, DOCX, XLSX, XLS, XLSM, XLSB, PPTX, TXT, CSV, PNG, JPG, JPEG, TIFF, TIF, BMP, WEBP. A file whose contents do not match its extension, or a password-protected Office file, is refused at upload with a reason, not discovered minutes later.
- Auto-detected document type: CIM, P&L, Balance Sheet, Cash Flow Statement, Tax Return, or Other. You confirm the detection; only confirmed statements unlock the financial modules.
- Scanned and photographed pages with no text layer are read optically and transcribed, and the transcription is marked as machine-read wherever it travels.
- Inline preview, multi-select delete, submit for analysis in bulk, replace a document with a newer version and reprocess it.
Documents
- DOCX
- PPTX
- TXT
Spreadsheets
- XLSX
- XLS
- XLSM
- XLSB
- CSV
Scans and photographs
- PNG
- JPG
- JPEG
- TIFF
- TIF
- BMP
- WEBP
Read optically when there is no text layer, and marked machine-read wherever the transcription travels
Detected on upload, confirmed by you
- CIM
- P&L
- Balance Sheet
- Cash Flow Statement
- Tax Return
- Other
Only confirmed statements unlock the financial modules
Refused at upload, with a reason
- A file whose contents do not match its extension
- A password-protected Office file
AI Insights extraction
- Revenue and EBITDA by year, gross and EBITDA margins, growth, a business description, risk flags.
- A confidence score per data point, computed from what the extraction found, identical on every screen that quotes it.
- Every extracted figure is cross-checked against the source table it came from. A history that disagrees with the document is withheld rather than shown, so you see a gap rather than a wrong year.
- A document in a different currency from the deal's is flagged and never silently converted.
- Periods are dated from the statement itself, including non-December year ends.
- A confidence score on the extraction, the same on every screen that quotes it
- Revenue and EBITDA by year, dated from the statement itself
Q&A
- Ask in plain language: "What are the top three customers by revenue?"
- Answers stream with inline markers; the Sources list shows only the documents the answer cites, numbered from [1] in the order cited. Click either to open the passage.
- An answer that cannot be supported says the excerpts do not show it, and cites nothing.
- Money reads in the deal's currency exactly as on every other screen. Rate any answer up or down; answer history is kept for 90 days.
- Load an industry playbook to run a structured question set.
Under the hood for the technical buyer
Documents are indexed for retrieval in Canada. For each question the relevant passages are retrieved and the model answers only from them, which is why every answer can cite its source. The Vera and AI governance page says where inference runs.
What comes out of it
A confirmed document set, scanned pages included; revenue, EBITDA and margins by year with a confidence score on each, ready for the Quality of Earnings and the valuation; and answers with numbered citations that open the passage.
Join the waitlist for the 14-day trial.
Sign-up is not open yet; we will email you when it opens.