AI Investment Due Diligence

Before You Write The Check

You can judge the founders and the market. What is harder to judge from a deck and a demo: is the technology real, is the build-vs-buy call right, can this team actually ship, and is the AI genuinely defensible or a thin wrapper on someone else's model. I stress-test exactly that, fast, and hand you a decision you can act on.

Book Assessment Call See The Method

The Problem

Demo day answers the wrong question. A polished demo tells you the product works once, in front of you. It does not tell you whether the AI is a defensible system or a call to someone else's API wrapped in a UI, whether the architecture holds at ten times the load, or whether the team can maintain what they built.

  • Is the model or pipeline actually theirs, or a thin wrapper on a foundation-model API with no moat
  • Does the data or workflow create a real advantage, or is it easily replicated
  • Is the build-vs-buy call right, or did they build what they should have bought (or the reverse)
  • Can this team ship and maintain what they built, or does the knowledge leave with one person
  • What breaks first under real usage, and what does fixing it cost

Visual Element

The Method: Technical Reality Check for AI Investments

The same structured due diligence used to evaluate technology acquisitions, extended with an AI-specific reality check. A fixed-scope sprint, not an open-ended engagement.

1

Tech Reality - What is actually built versus claimed. Architecture review against the pitch.

2

AI Reality Check - Is the AI real: data or model moat, training and inference economics, wrapper-or-not. The question generic technical due diligence does not ask.

3

Build vs. Buy - Was the right call made on what to build in-house versus license or integrate. Cost and risk of each path from here.

4

Team and Delivery Risk - Can this team ship and maintain it. Key-person risk, technical debt, what happens if the lead engineer leaves.

What You Get

A decision-grade report, not a technical audit nobody on your investment committee will read.

Green / Amber / Red Verdict

On each axis: tech reality, architecture and scalability, build-vs-buy, team capability, AI substance, key-person and tech-debt risk.

Five Questions for the Founder

The specific questions to put to the founder before you commit, and what a good versus evasive answer looks like.

AI-Reality Memo

Plain answer to the question your committee will ask anyway: is this AI, or an API call with a story. Data moat, model economics, and defensibility, argued with evidence.

30-Minute Debrief Call

Verdict walked through live, questions answered, before your decision deadline.

The Same Rigor, Proven

This method is the AI-specific extension of the rapid technical due diligence I already run for technology acquisitions - including a real case where it mattered.

  • A $200K real estate cleaning and scheduling platform acquisition: 2-hour structured assessment found the platform already covered about 80% of the needed functionality, quantified $220-345K of hidden infrastructure value, and returned a Buy recommendation with negotiation leverage
  • Full case study: see Rapid Technical Due Diligence
  • I am AI-native myself - I design and build the agentic systems I use daily, which is what lets me tell a defensible AI system from a wrapper instead of taking the pitch's word for it

Visual Element

Investment

Fixed fee, priced in line with market rates for early-stage technical due diligence. No hourly billing.

Rapid Screen

1-2 days. Go/no-go signal on tech reality, team, and whether the AI is real - for triage before a full look. From 2,000 EUR.

Full AI Investment DD

About 5 days. The complete report above, plus the debrief call. From 8,000 EUR.

Considering a larger technology acquisition instead of an early-stage investment?

See Rapid Technical Due Diligence - the same method scaled for company or platform acquisitions, from 5,000 EUR.

About To Back An AI Startup?

Book a 15-minute call to discuss scope and timeline before your decision deadline. Ex-CTO, 20+ years building software, and AI-native - I build the systems I evaluate.

Book Assessment Call