AI for commercial due diligence
Test the management case against the transcripts and the data room exports, and size the market bottom-up and top-down. The same evidence then carries into the cost case, the 100-day plan and the IC deck.
Harvest-thesis-test.docxExampleProject Harvest: Management Case Tested Against 24 Customer References
Assumption 1: 6% annual price increases hold through 2028
Management assumes 6% list price increases each year with no change in churn. Of the 24 references, 17 accepted the 2025 increase without pushback, but 5 of the 9 multi-site groups said a second increase above 4% would send them to tender. We rate this partly supported and model 4% for groups with three or more clinics.
Assumption 2: in-clinic share of blood panels rises from 41% to 60%
Management expects clinics to bring panels in-house as analyzer placements grow. Only 6 of the 24 references plan to add an analyzer, and 11 said reference lab pricing keeps routine panels sent out. The evidence does not support 60%, so our base case holds the in-clinic share at 45% by 2028.
CheckedEvery finding cites the reference number and minute mark. Ref 14 is a former employee, so it carries less weight.
How it works
- 1
Start from the deal folder
Pull in the data room files, the CIM and the management model, customer reference transcripts, expert call notes, client revenue files and last year’s report. The fonts and conventions already in the folder set the format it drafts in.
- 2
Ask for the piece of work
Run the thesis test or code the expert calls into themes. What comes back is a finished file, an Excel workbook with live formulas, a Word memo or a deck, and every figure names the transcript, tab or filing it came from.
- 3
Page-turn it before the IC
Read every file it wrote or edited, tie each figure back to its source and decide what survives. Nothing goes out before your review, so the call in the room stays yours.
Ask it for
“Go through the 24 customer reference transcripts for Project Harvest and flag every assumption the evidence does not support.”
Harvest-thesis-test.docx“Size the US veterinary diagnostics market from clinic counts and installed instruments, then reconcile it against top-down spend data and name each source.”
Vet-diagnostics-sizing.xlsx“Turn the Harvest findings into the IC storyline: one governing message, a 10 to 12 page spine and action titles that state each finding.”
Harvest-IC-storyline.pptx“Build the cost takeout case for Project Harvest from the HR and operations exports, with run-rate by lever, one-time costs and each lever’s assumption.”
Harvest-cost-case.xlsx“From the code review notes and the last 12 months of cloud bills, classify the tech DD findings as remediable or thesis-breaking, with a cost range on each.”
Harvest-tech-DD.docx“Rebuild NRR and the ARR waterfall from the customer-level revenue file and list every figure that does not tie to the April board pack.”
ARR-NRR-rebuild.xlsx
Inside the Deal Strategy & Value Creation mode
Executive storyline and decks
Titles every slide of the IC deck with the finding it carries and ghosts the exhibits to the house template, under one governing message.
Proposals and RFP responses
Turns a PE firm’s panel RFP into a compliance matrix and drafts each response section against the evaluation criteria and required forms.
Statement of work
Drafts the SOW for a diligence phase or value creation engagement, with deliverables, assumptions, exclusions and change control.
Project budget and cost forecast
Runs the monthly fee report on a live engagement, compares fee burn and percent complete to plan and restates the bottom-up estimate at completion.
Tie-out and page-turn review
Page-turns the report or banker book before it issues and traces every exhibit and figure to the file that supports it.
Commercial due diligence
Scopes the few questions that decide the deal and tests the management case against what customers and experts actually said.
Market sizing and evidence synthesis
Triangulates the market size, reconciles the bottom-up and top-down estimates, and codes the transcripts into themes with a verbatim quote behind each claim.
Value creation and transformation
Sizes the cost and revenue levers, builds the 100-day plan and tracks value capture against what the deal model promised.
Technology due diligence
Weighs architecture, tech debt, team and cloud cost, then writes the findings the IC rules on, each with a remediation cost range and a thesis verdict.
Operating metrics and reviews
Builds the quarterly operating review with the ARR waterfall, NRR and GRR by cohort, pipeline coverage and burn multiple.
- Office skills
- Excel, Word, PowerPoint, PDF
- Tools
- Read, edit and write files, Search files, Run code, Search and read the web, View images
- Connectors
- SharePoint, Google Drive, Gmail, and additional custom connectors
- House style
- Your fonts, layouts and rules, on every file
Who it’s for
Consultants who serve investors and acquirers across a deal, from commercial and operational diligence through integration, separation, value creation and exit, at strategy firms, deals practices and specialist boutiques, from analyst to partner.
Questions
Can we put data room documents into it?
Only when your NDA and your firm’s rules allow it, so check both before you move any export. Feeding confidential information to a third-party AI tool can count as a disclosure under a confidentiality agreement, and newer NDAs restrict AI use expressly. Ultramarine Work processes content on Paper Instruments’ servers, and deciding whether the exports can go in belongs to your client team, your firm’s risk function and counsel.
Will it invent a market size to fill the gap?
No, the skills in this mode tell it to cite the source file for every figure and to note what it could not find or reconcile instead of guessing. That is not a guarantee that every number is right. Re-derive the load-bearing math yourself and open each citation before the market size goes in front of the IC, because a sizing is only defensible when every input can be traced.
Can it run the expert calls for us?
It makes no calls, and nothing joins the line on your behalf. Hand it the transcripts and notes once the 30 expert calls are done. It tags the recurring themes and backs every claim with the exact words spoken. MNPI screening, consent to record and judging whether an expert is credible stay on your side of the desk, because tenure and recency are not things a document can settle.
Where does the target’s confidential data go?
Ultramarine Work changes files only in the project’s folder. When a request needs it, it can read files elsewhere on your computer, never passwords, keys or app data, and you can turn that off in settings. The model runs on Paper Instruments’ servers, so the content it works on is sent there to be processed. Paper Instruments is SOC 2 Type II, and our security and compliance details are on our compliance page.
Can we issue an AI-drafted report with our reliance letter?
Whether the report can carry your reliance letter is settled by your engagement letter and your counsel. Your name goes on the report only after you review every file it touched and run the tie-out. What a lender or buyer gets is whatever reliance the engagement grants. No regulator or professional body has endorsed the tool.