Question the data room instead of reading all of it
Upload contracts, filings, decks, spreadsheets and scanned reports. Ask across the whole set at once and get answers with the source passage attached. What you could not find is reported rather than glossed over.
Free plan. No card required.
What is due diligence software?
Due diligence software helps a team review a large volume of documents against a defined set of questions, and evidence its conclusions. The hard part is never storage. It is being able to ask a question across everything at once, get an answer you can trace to a page, and know what the material did not cover.
The specifics
Every format the data room actually contains
Portable documents, spreadsheets, word processing files and scanned images. Scanned pages go through document-native optical character recognition rather than a general vision model, because a general model occasionally reformats a table and a reformatted balance sheet is the worst failure available here.
Ask across the whole set
One question, every document, answers with the file and the passage attached. Retrieval combines keyword and meaning-based search, then re-ranks, so the right passage surfaces even when the wording differs.
Red flags, with the evidence
Findings are reported with the text that produced them, so a flag can be judged rather than taken on trust.
Financial statements out of a filed table
Statements in a document, including older filings and reports that were never tagged, parsed into the same concepts as filed data, with the units declaration read explicitly rather than assumed.
More on thisTranscript and call analysis
Audio transcribed with timestamps and ingested like any other document, so a retrieved passage points at a moment in the recording.
A workspace per deal
Documents, questions, findings and notes scoped to one process, exportable when it closes.
How the work actually goes
- 01
Load the room
Drop in everything. Scanned material is read, spreadsheets are parsed, and each file is chunked and indexed for retrieval.
- 02
Work your question list
Ask the standard set first, then the ones this deal raises. Answers arrive with the file, page and passage.
- 03
Check what is missing
An absent answer is reported as absent. A tool that quietly produces a plausible answer to a question the room does not address is worse than one that produces nothing.
- 04
Write it up
Findings and citations export into a memo or a committee pack rather than being retyped.
An uploaded document is untrusted text, and it is handled as such
Anything the system reads that neither we nor you wrote arrives in the same channel as the instructions: uploaded files, scanned images, filings, transcripts. That is an injection surface, and reading documents is the feature that creates it. Third-party text is fenced with an explicit boundary, scanned deterministically for attempts to steer the answer, and any attempt is surfaced with the offending words quoted. That mitigates the risk. It does not eliminate it, and we would rather write that down than imply otherwise.
Frequently asked questions
What file types can I upload?
Portable documents, spreadsheets, word processing documents, plain text and scanned images. Scanned material is read with optical character recognition. Audio can be transcribed and ingested the same way, with timestamps preserved so a citation points at a place in the recording.
Is there a file size limit?
Four megabytes per upload. That is the honest number rather than an aspirational one: the platform this runs on rejects a larger request body before our own code sees it, so a higher advertised limit would be a promise broken by an opaque error.
Can it read scanned documents?
Yes, using a document-native model rather than a general vision model. The distinction matters because a general model asked to read a balance sheet will sometimes paraphrase or reformat the table, which is precisely the failure mode you cannot afford when the output is financial data.
Who can see the documents I upload?
They are scoped to your account. Deleting a document removes it from the database and from the search index, which are two different stores, and account deletion purges both.
Try it on a document you know
Upload something you have already read and ask it the awkward question.