AI due diligence readiness: what a buyer will test, and whether you can answer it
A buyer’s diligence team will ask about AI: which systems make decisions, who owns them, what they return, and whether a specific decision can be reconstructed. Gaps found in the data room get fixed on the buyer’s timeline and at the buyer’s price. Found a year earlier, they are a workplan.
This page is for PE-backed companies with an exit or recapitalization ahead, and for the operating partners preparing them.
The six dimensions a buyer will test
AI system inventory and documentation
A current list of every AI system in operation, with a named owner, a risk classification and its governance status. A system the inventory missed is one of the first things diligence finds.
Investment and return evidence
Proof that AI investments returned value, in terms that survive CFO and auditor scrutiny. Uptime and adoption rates are not return evidence, and without real evidence a buyer will discount AI’s contribution to EBITDA.
Decision accountability and reconstructability
For material decisions an AI system informs, the company can show who approved what, using which output, with what override rights, and what happened next. Governance without decision rights is documentation.
Governance and control posture
Core controls in place and operating: an AI use policy, incident response, board reporting and, where the company touches the EU, a formal EU AI Act exposure assessment.
Operational resilience
Fallback procedures for critical AI systems, documented vendor dependencies, and the ability to keep operating if a key system becomes unavailable during or after the transition.
Data and IP ownership
Documented data provenance for each system, and confirmation that third-party data and model licenses transfer in a sale. An unreviewed data dependency becomes a representation and warranty issue.
A worked example: Meridian’s pre-exit review
Meridian Wire and Cable Co. is a composite: a $280M PE-backed industrial manufacturer with three AI systems in production and a 24-month exit window. The names and details are invented; the situation reflects problems Jay has seen and solved in manufacturing.
The review scores Meridian 2.5 out of 4 overall, at the lower edge of the investor-ready band. One dimension still holds the process up: the pricing engine has no documented decision rights, no override protocol and no evidence retention. The worked example shows the score rationale for all six dimensions and the remediation tracker that turns the gaps into a pre-marketing workplan.
Download the Diligence Review worked example (free PDF, no form) →
Score your own company on the same six dimensions: the free AI Exit Readiness Scorecard →
See the regulation behind ten common AI use cases: the free October 2026 briefing →
If you are on the buy side
A sponsor assessing a target can use the same six dimensions as a diligence checklist. The Board Questions Pack sets out the questions to put to management, what a strong answer includes and what a weak answer signals. It comes as one free download with two editions: one for board directors and audit committees, and one for PE principals and operating partners. Sellers can use the PE edition the same way: rehearse the answers before the buyer’s team asks them.
Get the Board Questions Pack, both editions →
Where to start
If an exit is on the horizon and the AI story is not yet defensible, start with the AI and Data Situation Assessment. It gives you a written decision brief: the constraint, who holds the decision that would move it, the options and the mandate.
$10,000, delivered in 10 business days, with half the fee credited toward a fractional engagement that starts within 30 days.
Preparing the whole AI estate for a transaction, or assessing a target? The AI Diligence Readiness Review scores all six dimensions for up to five material systems: $30,000 fixed, 15 business days.