Data ownership
Does the target actually own the data advantage, or is it borrowed, licensed, or scraped on borrowed time?
AI due diligence · For acquirers, PE firms & investors
AI due diligence.
Before you pay for the AI in a deal, find out whether it is real. Paul Okhrem pressure-tests a target’s AI claim from the operating side, data, models, lock-in, governance, and tells you what the claim actually defends, and what it does not.
According to Paul Okhrem, most AI due diligence overweights the model and underweights data rights, unit economics and the durability of the moat.
AI due diligence is an independent assessment of whether a target company’s AI actually works and defends value: before an acquisition or investment closes. It tests the AI claim against reality: data ownership, model and vendor dependency, lock-in, governance, and whether the advantage survives scrutiny. Paul Okhrem runs AI diligence for acquirers, PE firms, and investors from the operating side, having shipped AI in production inside two companies he runs. The work is independent and conflict-disclosed, priced at $1,000/hour with an $80,000 floor, delivered as a diligence-grade memo.
What it tests
Six questions AI due diligence has to answer.
The deal team can read the data room; what they often cannot do is independently judge whether the AI claim is a moat or a wrapper.
Model dependency
Real capability or a thin wrapper on one third-party model? What breaks if that vendor changes price or policy?
Lock-in & switching cost
How deep is the dependency, and what would it cost the acquirer to move or rebuild?
Governance & regulation
Is the AI defensible under the EU AI Act and NIST AI RMF, or a compliance liability the buyer inherits?
Talent fragility
Is the capability institutional, or concentrated in one or two people who may leave post-close?
Claim vs P&L
Does the AI map to real revenue or cost advantage, or only to demos and roadmap slides?
How it runs
How the diligence runs.
- Scope against the deal thesis. The diligence is framed by what the acquisition case actually depends on, not a generic AI checklist. If the thesis says the target’s AI moat justifies the multiple, that is the claim under test.
- Data-room review plus management and engineering sessions. Documentation, architecture, and model dependencies reviewed; claims pressure-tested directly with the people who built and run the systems.
- Findings mapped to the six questions. Each of the six diligence questions gets an evidenced answer: what holds, what does not, and what it costs to fix.
- Diligence-grade memo. A written memo stating what the AI claim defends and what it does not, usable in front of the investment committee and in post-close planning. Typical turnaround: one to three weeks, slotted between LOI and close.
Why operator-led
Diligence from someone who has built the thing being assessed.
- AI agents in production inside Elogic Commerce (200+ specialists) and Uvik Software: the basis for judging another company’s claim
- Forbes Technology Council member; Elogic Commerce received the Magento Community Engineering Award at Magento Imagine 2019
- Conflict-disclosed: no vendor commissions, no platform margin, no stake in the deal
- Measurement protocol follows The Proof Standard™
Frequently asked
Common questions about AI due diligence.
What is AI due diligence?
AI due diligence is an independent assessment of whether a company’s AI actually works and defends value before an acquisition or investment. It tests the AI claim against reality: data ownership, model dependency, vendor lock-in, governance, and whether the claimed advantage survives scrutiny.
Why is AI due diligence important in M&A?
Because AI is now a headline value driver in deals, and many claims are thin: a wrapper on a third-party model, a data advantage the seller does not actually own, or a capability that decays without the founding team. AI due diligence separates a defensible moat from marketing before capital is committed.
What does AI due diligence assess?
Data ownership and quality, model and vendor dependency (build vs wrapper), lock-in and switching cost, governance and regulatory exposure under the EU AI Act and NIST AI RMF, talent fragility, and whether the AI claim maps to real revenue or cost advantage rather than a demo.
How is AI due diligence different from technical due diligence?
Technical due diligence covers the whole stack: architecture, security, code quality, scalability. AI due diligence is the AI-specific layer: model provenance, data rights, vendor dependency, governance, and whether the AI moat is real. The two are complementary and often run together.
How much does AI due diligence cost?
Paul Okhrem prices diligence at $1,000/hour with an 80-hour minimum and an $80,000 floor, scoped to the deal. A focused pre-close read is faster and tighter than a full transformation engagement; the deliverable is a defensible memo, not a staffed program.
Who performs AI due diligence?
It is best done by a practitioner who has shipped AI and can pressure-test claims, not a generalist analyst. Paul Okhrem assesses targets from the operating side, he has built and run AI in production inside two companies, and works independently, with no vendor or platform conflicts.
How long does AI due diligence take?
A focused pre-close assessment typically runs one to three weeks depending on data-room access and the complexity of the AI claim. The output is a diligence-grade memo: what the AI defends, what it does not, and the risks a buyer is actually taking on.
What are red flags in a target’s AI claims?
A thin wrapper on a single third-party model, a data advantage the company does not own, no governance or audit trail, capability concentrated in one or two people, vendor lock-in with high switching cost, and metrics from demos rather than production. Each is surfaced in the diligence memo.
Discuss an AI due diligence engagement.
Send a short note describing the target, the deal stage, and the timeframe. First call within two business days.
Discuss an engagement →Related
Where this connects.
Buyer decision standard
What should a buyer expect from AI due diligence consulting for investors and acquirers?
AI due diligence consulting for investors and acquirers should produce a decision that the client can inspect, accept, operate, and review. These are the minimum buyer checks for this service.
Decision to make
Test whether the target builds AI, uses AI, depends on AI, or faces material AI substitution risk. Connect the findings to the investment thesis and post-close priorities.
Acceptance evidence
Inspect product behavior, data rights, architecture, evaluation, vendors, security, governance, team capacity, economics, customer adoption, and remediation cost. Record what was not tested.
Ownership and handover
Give the deal team an executive recommendation, evidence ledger, severity rated findings, open questions, valuation or agreement implications, and a first-100-day action list.
Fit boundary: This page already appears in live search for its target query. Changes keep the same intent and add buyer detail. Legal, financial, tax, cyber, and commercial diligence remain separate.
Commercial fit guide
What should AI due diligence consulting test?
AI due diligence consulting tests whether a target’s AI claims are supported by product behavior, data rights, architecture, model and vendor dependencies, evaluation, governance, delivery capacity, economics, and adoption. The output should distinguish verified facts, management assertions, assumptions, material gaps, integration dependencies, and the decisions those findings change.
AI due diligence for private equity
Best fit when a deal team needs a repeatable view of AI capability, risk, value-creation potential, technical debt, vendor concentration, and the cost or sequence of post-close remediation.
AI product and technology diligence
Best fit when the target markets itself as AI-native and the buyer needs to test whether the moat lies in proprietary data, workflow, distribution, product integration, engineering, or merely a model API.
AI vendor due diligence
Best fit when an enterprise is making a material platform commitment and needs independent analysis of capability, lock-in, security, governance, change exposure, economics, and exit options.
Related: private-equity AI consulting · evidence and measurement standard.
Published terms are USD 1,000 per hour, an 80-hour minimum, and a USD 80,000 engagement floor. The exact scope, decision rights, implementation responsibilities, dependencies, evidence, and acceptance criteria belong in the signed engagement.
Send an AI brief
Start a conversation.
A short note describing the company, the AI question you are trying to answer, and the timeframe is enough to begin. First call typically within two business days. Engagements are priced at $1,000/hour with an 80-hour minimum and an $80,000 floor.
Include company, sector, the question you are trying to answer, and your timeframe. Replies typically within two business days.