Model, moat or wrapper?

[ MSP.DD.01 ] The four questions

Every report answers four questions.

[ MSP.DD.02 ] Why it matters

AI claims are in every deck. The risks are in the code.

A standard technical due diligence with a specialist firm costs $35,000–95,000 and takes two to four weeks, so on smaller rounds it often gets skipped. That is where AI risk goes unpriced.

[ MSP.DD.03 ] The offer

Fixed price. Fixed scope. Five working days.

[ MSP.DD.04 ] How it runs

Five working days from access to readout.

1. Day 1

   Access and kickoff

   Read-only repository access under NDA, the technical documents in the data room, and the questions that matter for this deal.

2. Days 2–3

   Evidence

   Scans of code, dependencies, secrets and licences. The architecture map: what's built, what's rented. The cost model per customer.

3. Day 4

   Interviews

   The CTO and the product lead. Every answer is checked against the evidence from days 2 and 3.

4. Day 5

   Report and readout

   A verdict per area, the evidence behind it, and the issues to raise in negotiation. A 60-minute readout with your deal team.

We build AI systems for a living, so we know where they break. We read the code, not a checklist. We don't give legal or financial advice and the report isn't a certification: issues that need a lawyer go to yours.

[ MSP.DD.05 ] Questions investors ask

The questions, answered.

[ MSP.DD.06 ] FAQ

Before you ask.

What is AI due diligence?

A review of an AI company before you invest in it or buy it. It answers four questions: what the company built versus what it rents from model providers, what the code hides, what the product costs to run, and what the EU AI Act will ask of it. It sits next to financial and legal due diligence and covers what they don't.

How long does it take?

Five working days from the day we get access. The report gives a red, amber or green verdict per area, the evidence behind it and the issues to raise in negotiation, followed by a 60-minute readout.

What does it cost?

From €7,500, fixed before we start. Published ranges for a standard technical due diligence with a specialist firm run from $35,000 to $95,000 over two to four weeks.

What access do you need?

Read-only access to the code repositories under NDA, two interviews (usually the CTO and the product lead), and the technical documents in the data room. If the company won't share code, we work from interviews and documents and say so in the report.

What happens to the company's code?

Access is read-only. Code is stored in the EU, deleted 30 days after delivery, and never used to train models.

How is this different from a standard technical due diligence?

A standard review covers architecture, scalability and the team. We cover those too, and add the AI-specific questions: how much is rented from model providers, how much code was AI-written and what that left behind, model cost per customer, and the EU AI Act. And it takes five working days, not two to four weeks.

Do you give legal advice or certify anything?

No. We flag issues for your lawyers, such as licence conflicts or an AI Act classification question. The report is evidence and judgement, not legal advice, financial advice or a certification.

Can founders buy it?

Yes. The Founder Readiness Check runs the same analysis on your own code before a round, from €3,500, with a fix list ranked by what an investor would flag first.

What does the free scan check?

For a public GitHub repository: how many of the last 100 commits carry AI-tool markers, which AI providers the code depends on, known vulnerabilities in its dependencies, committed environment files, and how concentrated the authorship is. It reads public data only.