Diligence what the data room cannot show you.
Logos, case studies and reference calls all originate with the company. The Index records which of a vendor's published customer claims an enterprise has independently confirmed, how far each deployment has actually gone, and which of them did not hold.
A signed contract and a live deployment are different assets
One renews because it works. The other renews because nobody has looked.
Index edition 2026.4 · compiled 18 Oct 2026
Three things a data room is not built to establish
Diligence confirms that revenue was contracted and recognized. It is not designed to confirm that the software behind that revenue is running inside the customer.
A logo establishes a relationship
It does not say which business unit, which function, how many users, or whether anything went past the pilot. Logo walls rarely distinguish a global rollout from a six-week trial in one team.
ARR does not tell you what is running
A contract that renews because the deployment is load-bearing and one that renews because nobody has reviewed it look identical on a revenue schedule, and behave very differently at the first budget cycle under pressure.
The reference list is curated
You speak to the sponsor who staked something on the deployment succeeding. You do not hear from the two accounts where it stalled, and the company is not obliged to mention them.
Every claim a company makes, with what each one survived
Across a single target or across the whole portfolio, in the same structure: the claim, the level it reached, the deployment stage, and the date anyone last checked.
| Holding | Category | Claims | L2 | L3 | Deepest stage established | Currency |
|---|---|---|---|---|---|---|
| XYZ AI | Claims and servicing automation | 13 | 3 | 8 | Full deployment | Current |
| Company B | Document processing, trade finance | 22 | 9 | 6 | Partial deployment | Current |
| Company C | Clinical documentation | 31 | 4 | 11 | Full deployment | Renewal due |
| Company D | Demand forecasting | 46 | 6 | 2 | Yet to start | Unconfirmed |
| Company E | Network fault resolution | 18 | 5 | 9 | Full deployment | Current |
| Company F | Agentic customer operations | 54 | 2 | 1 | Yet to start | Unconfirmed |
Fifty-four published claims and one confirmed deployment is a finding. So is a holding whose deepest established stage has not moved in two review cycles.
What we establish, and what remains yours to judge
The Index produces facts about deployment. It produces no view on the company, no score, no rank and no recommendation, and it never will.
- Whether each published customer claim is real
- Whether a commercial relationship exists behind it
- Whether the organization says it is running
- How far it has reached inside that organization
- The date every one of those was last confirmed
- Value a company or opine on a price
- Score, rank or place vendors on a quadrant
- Recommend, endorse or advise against an investment
- Disclose the identity of any enterprise that answered
- Hold back a determination because of who commissioned it
This is deliberate. An institution that both establishes facts and ranks the companies those facts concern eventually has an interest in the ranking. We took that option away from ourselves so that the determinations can be relied on by people on both sides of a trade.
Four points in an investment process where deployment evidence changes the number
Thesis and category
Before the first meeting. How much of a category has any confirmed deployment behind it at all, which use cases have working implementations, and which are still entirely narrative.
Tells you whether the category is real yetPre-investment
Every customer claim the target publishes, taken through the five layers. Claims that did not hold are reported alongside the ones that did, and so are the ones nobody answered.
Separates contracted revenue from running softwareInvestment committee
External, dated evidence that did not come from the company, attached to the paper. The method is published, so the determination can be interrogated rather than taken on faith.
Gives the committee something it can argue withPortfolio and re-rate
Annually across holdings. Which deployments are still confirmed, which lapsed into unconfirmed, and where the deepest established stage stopped moving.
An early signal, a full cycle before churn appearsThree ways funds use the Forum
The method, the fee and the published outcome are identical whoever commissions the work. That is the point of commissioning it.
Pre-investment claim verification
Every customer claim a target has published, run through the method inside a defined window, with the full log for each determination.
- Three to six weeks
- Scoped by claim volume
- You are never named
Portfolio deployment review
Every holding, once a year, in one structure. What is still established, what lapsed, and where a company's deepest confirmed stage has not moved.
- Annual cycle
- Priced per holding
- Briefed to the partnership
Category depth before a thesis
One category, every vendor in it, every claim across all of them. Where the working deployments actually sit, and which use cases have none behind them anywhere.
- Six to eight weeks
- Written and presented
- Underlying data included
Determinations publish. Nothing about the commission does. Entries enter the record on the Forum's own cycle rather than on your timetable, carry no reference to why they were made, and never name the party who asked. A live process is not inferable from a record.
The whole register, not one record at a time
Individual records are public and free to anyone holding the link. Funds that work across a category take register access: search every vendor and every claim, filter by level, stage, industry and currency, export the underlying data, and receive the weekly record of what was established, what lapsed and what did not hold.
Looking at a company we have not indexed yet?
Tell us who. If the vendor is not yet in the Index we open a record, capture every customer claim standing on their own site, and publish what is there. Capture is free, nobody is asked for permission, and the company finds out the same way everyone else does.
What funds ask us first
They will see entries appear on their record, which happens continuously across every category whether or not anyone commissioned anything. They will not see who asked, when it was commissioned, or that a commission existed. We hold that permanently, and it is not disclosed to the company, to another fund, or to a counterparty.
Entries publish on the Forum's own cycle rather than on a deal timetable, and they read exactly like every other entry in the Index. There is no commissioning stamp, no cluster of same-day publications, and no pattern that distinguishes an investor-commissioned determination from a vendor-commissioned one.
No, and this is a standing constraint rather than a current position. We establish whether deployments are real and how far they have gone. We do not score, rank, value, endorse or advise, because an institution that does both eventually has an interest in the second one.
Then that is the finding, and it is worth knowing before a committee rather than after. Where a company has published little, we can work from the customer references supplied in the process instead, with the same method and the same publication rules applied to whatever we are given.
You receive the full log for every claim: what was captured and when, what was inspected at the vendor, when the enquiry went to the organization, what came back and on what date. What you never receive is the identity of the organization, which is the condition on which they answer at all.
Every determination carries the date it was established and runs current for twelve months, after which it reads unconfirmed. The level itself never expires. For a holding you are monitoring, a lapse into unconfirmed is often the first visible signal that something stopped being looked after.
Check the deployment before the committee checks the model
Commission a pre-investment verification, a portfolio review, or category depth before the thesis is written. The method, the fee and the published record are the same either way.