The data room closes Friday.
Forty targets, a few thousand documents, and the same twelve questions to answer about every one of them. Ragextract reads the room once and returns the comparison — each figure citing the document and the page it came from, including which filing superseded which.
Free credits on sign-up. No card, no subscription.
| Row | Documents | Revenue growth | Valuation multiple | Customer risks | ESG disclosed |
|---|---|---|---|---|---|
| Kestrel — IM.pdfQ3 trading update.pdfManagement accounts.xlsx | 22.6% Trading update, p.2 | 11.4× EV/EBITDA IM, p.18 | Top customer 34% of revenue; contract expires 2027 IM, p.51 | Yes IM, p.66 | |
| Harlow Group — IM.pdfManagement accounts.xlsx | 5.2% Accounts, p.9 | 7.1× EV/EBITDA IM, p.12 | Top 5 customers 71% of revenue; no long-term contracts IM, p.33 | Not stated | |
| Pelham Industrial — IM.pdfESG report.pdf | 9.8% IM, p.7 | 9.0× EV/EBITDA, 1.8× EV/Revenue IM, p.21 | Two customers over 15%; both renewed 2026 IM, p.40 | Yes ESG report, p.3 | |
| Ashcombe — teaser.pdf | Not stated | Not stated | Not stated | No Teaser, p.2 |
A number you cannot source is a number you cannot use.
An investment committee paper is worth exactly as much as the footnote under it. A growth rate with no page reference is not a finding; it is something to go and check, which means the work has not actually been done.
Every cell carries the document and page it was read from, plus a confidence score, so the review is targeted at the handful of answers that look thin instead of being a second pass over the room. When a figure is challenged in the meeting, the source is one click away rather than one afternoon away.
| Document | Spread | Leverage covenant |
|---|---|---|
| Revolver +1 | SOFR + 300bps Am. 1, p.2 | 3.75× net, springing at 35% drawn p.62 |
One subject, many filings.
A target is not a document. Its information memorandum, its management accounts and the trading update that overtakes both are one subject, and a table that reads them separately will report a figure the target itself has already revised.
Group them into a bundle and the bundle is the row. Order carries precedence, so putting the most authoritative document first is how “most recently reported” comes to mean what you intended. A bundle holds up to fifteen documents, and a tight one answers better than a large one — the first three are included in the column’s rate, and each past the third is read separately.
Where it fits.
The first decision is always what a row is. A row is the thing you are comparing — which is a company more often than it is a file.
| The work | What a row is | Start from |
|---|---|---|
| Diligence in a data room | A target, bundled from its filings | |
| Credit agreement review | An agreement, with its amendments | |
| Earnings and transcript work | One company, one quarter | |
| Portfolio monitoring | A portfolio company |
The presets ship in the application and their prompts are illustrative starting points rather than financial definitions — they are meant to be edited. Diligence in a data room and credit agreements are both worked through end to end, including what tends to go wrong.
Every row here is a company. Once a deal is live, the work moves inside one target’s paper — which contracts need consent, which let a counterparty walk — and a row becomes a contract instead. That is the deal teams page.
For documents that keep arriving.
Monitoring is not a project with an end date. Compliance certificates, quarterly packs and amendments land on their own schedule, and the table is out of date the moment one does.
Put a table in auto mode and it runs itself against every new document added to the workspace, so the comparison is current without anyone remembering to press anything. Corrections survive it: type over a cell and the original answer is kept underneath, with who changed it and when.
Reading is charged once.
A workspace is billed for the pages it reads, not for the questions asked of them. The question that comes up in the committee meeting costs the same as the ones you planned — which is what makes it worth asking rather than noting for next time.
| A 40-agreement review | Volume | Cost |
|---|---|---|
| 40 credit agreements, 120 pages each | 4,800 pages | 4,800 credits |
| 4 questions asked of all 40 | 160 cells | 160 credits |
| A covenant question added at the IC | 40 cells | 40 credits |
Start with one target.
Upload a single company’s filings, bundle them, write two columns and read every citation. One row is enough to tell whether the answers survive your documents, and that is the only test worth running before the room opens.
Nothing on this page is investment advice, and the presets are not financial definitions.