AI Due Diligence for investors
Whether the target’s AI is real, what it costs to keep running, and what you are actually buying.
Built for private equity and venture funds, family offices and M&A boutiques. It is transaction work, not consulting work: it runs on your timeline, it produces a verdict, and it is over.
Three levels, nested
| Level | Scope | Typical use |
|---|---|---|
| 1 – Red-flag screen | Three to five days. An eight-signal gate returning GO, GO WITH FOCUS, or STOP. | Before the letter of intent, when access is limited and a full diligence is not yet justified. |
| 2 – Standard | Full workstream coverage with scoring and evidence. | Confirmatory diligence, where AI is material to the thesis. |
| 3 – Deep | Extended evidence, management testing, cohort assessment of the target’s team. | Large tickets, or when AI is the thesis rather than a feature. |
The Level 1 fee is credited in full against Level 2. If the screen says stop, you spent a few days to avoid a few weeks – and that is a good trade even when the deal was fine.
What gets examined
- Whether the AI is real or a wrapper with a story – tested, not asked.
- Who actually holds the rights to the data the models were trained on, and what happens to those rights at closing.
- Key-person risk: what walks out of the building when three people leave.
- The whole AI estate, not only generative models – classical machine learning, deep learning, computer vision, video, audio, forecasting. The economics here are not token economics.
- The embedded AI inside the target’s SaaS stack: what is switched on, whether it can be switched off, who controls updates, what it costs, and where the data goes.
- Unit economics at plausibility level: does it make money at volume, or only in the deck.
The deliverable is a verdict, not a method
One page with the exit you can act on, the short list of what to verify if you continue, and the evidence behind each finding. The scorecard sits underneath and you get all of it – but you should not have to read the method to know what to do.
The bridge to management quality
Where it matters, the maturity diagnostic runs on the target’s own executive team as evidence inside the diligence, graded against the same scale. A management team that cannot agree on where it stands today will not execute the AI plan it is being valued on.