AI Due Diligence for Investors
Separate real AI capability from pitch-deck fiction
Before you commit capital, you need independent verification. Is the AI real? Can it scale? What are the technical risks? We provide investors and acquirers with rigorous technical assessment of AI companies and AI-dependent businesses.
Red Flags We Look For
Common issues we find in AI due diligence
Demo ≠ Product
Impressive demos that rely on manual processes, hardcoded rules, or cherry-picked examples. The gap between demo and production-ready AI.
Data Dependency Risk
Models trained on data the company doesn't own, can't reproduce, or may lose access to. Single-source data dependencies.
Scaling Cliff
Architectures that work at demo scale but break at 10x or 100x. Missing infrastructure for production workloads.
Key Person Risk
All AI knowledge concentrated in one or two people. No documentation, no reproducible training pipelines, no institutional knowledge.
Compliance Gaps
GDPR, AI Act, or industry-specific regulations not addressed. Processing personal data without proper legal basis or safeguards.
Vendor Lock-in
Critical dependency on a single AI provider (OpenAI, Google, etc.) with no abstraction layer or migration path.
Quick Answers
How much does AI due diligence cost?
Our AI technical due diligence starts at CHF 15,000 for a focused assessment. Comprehensive evaluations for larger AI companies or multi-product portfolios range from CHF 25,000–50,000. The investment typically represents less than 0.1% of the deal value while protecting against material technical risks.
Can you work within deal timelines?
Yes. Our standard turnaround is 2–3 weeks from kickoff to final report. For time-critical deals, we offer an accelerated 7–10 day assessment that covers the highest-risk areas. We've worked within tight M&A timelines for deals ranging from Series A to nine-figure acquisitions.
Do you need access to source code?
We prefer architecture-level access: system designs, data flow diagrams, and guided code walkthroughs with the technical team. Full source code access isn't required but can enable deeper assessment. We work under strict NDA and can accommodate clean-room arrangements.
How do you handle confidentiality?
Swiss jurisdiction provides strong legal protections. We sign comprehensive NDAs before engagement, conduct assessments under strict information barriers, and deliver reports through secure channels. We never share findings with competing deals or third parties.
What if the target company is uncooperative?
We can conduct outside-in assessments using public information, published APIs, and product testing. While less comprehensive, this still identifies architectural patterns, technology choices, and potential risks. We clearly flag what we could and couldn't verify in our report.
What We Evaluate
AI Capabilities Verification
Independent testing of AI claims. We design evaluation protocols, test on representative data, and benchmark against alternatives to verify the company's AI actually works as advertised.
Technical Architecture Review
Assessment of ML infrastructure, model serving, data pipelines, and system design. Can this architecture handle production scale? What's the tech debt situation?
Data Asset Evaluation
Quality, provenance, and defensibility of training data. Data collection sustainability, competitive moat from data, and licensing/IP risks.
Team & Process Assessment
ML engineering maturity: version control, experiment tracking, CI/CD for models, monitoring, and incident response. Does the team have depth?
Regulatory & Compliance Posture
GDPR readiness, EU AI Act classification, industry regulations. Current compliance state and cost-to-comply estimates.
Scalability & Unit Economics
Inference costs at scale, compute requirements for retraining, infrastructure costs trajectory. Does the AI economics work at target scale?
Due Diligence Process
Structured assessment in 2–3 weeks
Scoping & Document Review
Review pitch materials, technical documentation, architecture diagrams. Define assessment scope and key questions from the investor's perspective.
Technical Deep Dive
Architecture review, code walkthrough with the team, data pipeline assessment. Independent testing of AI capabilities against claims.
Risk Analysis & Benchmarking
Identify technical risks, score by impact and likelihood. Compare capabilities against market alternatives and state-of-the-art.
Investor Report Delivery
Comprehensive report with executive summary, detailed findings, risk matrix, and investment recommendation framework.
Investor Report Contents
Executive Summary
- Go/No-go recommendation with confidence level
- Top 5 risks ranked by investment impact
- Key strengths and competitive advantages
- Comparison to market alternatives
Technical Assessment
- AI capabilities verification results
- Architecture maturity scoring
- Data asset quality and defensibility analysis
- Scalability projections and bottlenecks
Risk & Valuation Impact
- Technical risk matrix with probability/impact
- Cost-to-fix estimates for identified issues
- Compliance cost projections (GDPR, AI Act)
- 12-month technical roadmap assessment
Who Uses Our AI Due Diligence
The Swiss Advantage in Due Diligence
Switzerland's neutrality and data protection standards make it the ideal base for independent AI assessment. No conflicts of interest with US or Chinese AI vendors. No pressure to greenlight deals. Just rigorous, independent technical truth.
Invest with confidence, not hope
67% of the AI systems we assess have material technical risks that weren't in the pitch deck. Don't let your next AI investment become an expensive lesson.
Schedule Due Diligence