The number gets set while the room is still half-read
Traditional diligence samples the room and reports weeks later. By then the price is anchored and the findings are too late.
Whoever finds it first frames it
Whoever surfaces an issue first controls how it gets framed for the rest of the negotiation.
One dimension isn’t diligence
A quality-of-earnings read covers one dimension. The gaps show up between disciplines.
A risk list is not an argument
Most diligence is a static risk list. Nobody hunts for value-creation levers.
A sample is not the room
Coverage gaps are how deals go wrong after closing.
A finding isn’t finished until it can be traced
Every claim cites a specific document and passage. No citation, no finding.
The schema enforces it. We publish the contract instead of a sample, because client work is confidential.
A tested pipeline, run the same way every time
Anyone can point an agentic tool at a data room. The question is what happens every time, by design.
- Fabricated or hallucinated citations
- Privacy leaks
- Scope refusals
- Drafts below the reviewer threshold
7 hard-block conditions in total. Every one routes to a person.
- 01
Grounded retrieval
Each module dispatches its own sub-agents at your documents.
- 02
Drafting
The schema rejects any claim without a document and a passage behind it.
- 03
Generate-critique-retry
A reviewer agent critiques the draft and sends it back.
- 04
Four-layer grounding check
Index, quote, entailment, second opinion.
- 05
Human approval gate
One mandatory human step. It runs every time.
Built into how the pipeline runs
Each point below is a mechanism. None of them is a certification.
- EU-hosted analysis
- All analysis runs on Google Vertex AI in an EU region, under a Data Processing Agreement.
- Zero retention at the model provider
- The model provider does not keep your documents or train on them.
- Pseudonymized before processing
- Confidential data is pseudonymized before analysis. Reversed only in the final output.
- One accountable point of contact
- A named person reviews every output before it reaches you.
31 modules, 254 sub-agents, 5 waves
Factum is already built. Every module is a team of specialists with a pre-scoped brief.
Counted from the product source. The largest module fields 21 sub-agents.
One per discipline or deliverable.
Dispatched across 5 dependency-ordered waves.
A full 11-discipline pass, produced in days. Human review comes after it.
The Diligence Sprint
All 11 disciplines in one engagement, plus three synthesis layers. Run it on a target, a company you hold, or your own business.
- Financial
- Commercial
- Legal
- Tax
- HR
- IT
- ESG
- Vendor
- Operational
- Valuation
- Insurance
Three material findings your own team hadn’t already identified, or the engagement is free.
For pilot mandates. We carry the risk of this being early, not you.
One evidence base, five moments
Nothing gets rebuilt when the next moment arrives.
01Acquiring
From IM screening to closing, while the findings can still move price and structure.
02Financing
The memo a bank or investor underwrites from, built off the same evidence base.
03Holding and building value
Covenant headroom, KPI drift, early warning signals. Plus the value-creation levers a risk list ignores.
04Selling
Data-room quality, financial presentation, management strength, equity story. The buyer’s pass, run first on your side.
05Restructuring
Pre-IBR and WHOA: going concern, creditor position, covenant headroom. While the options are still open.
WHOA and IBR work is bound to Dutch jurisdiction. The rest of the cycle is jurisdiction-independent.