Technical Due Diligence & AI Claim Verification
Before you buy, fund, or rely on an AI system, we independently test what it can actually do.
Who commissions this
A material decision needs technical evidence.
For enterprise procurement teams, technical leaders, investors, and acquirers evaluating a material AI purchase or investment.
Evaluation scope
What we examine.
- Reproduce material claims against explicit baselines on representative customer data.
- Inspect evaluation leakage, contamination, test-set selection, and unrealistic assumptions.
- Measure human intervention, failure rates, latency, throughput, and operating costs under agreed conditions.
- Examine architecture, model dependencies, integration constraints, and production readiness.
Representative deliverables
Evidence you can review and use.
- Claim register: evidence, independent result, confidence, and limitations
- Reproducible test protocol and evaluation artifacts
- Technical risk register and unresolved evidence requests
- Decision brief: supported, conditional, unsupported, and untested claims
Decision outcome
A defined decision outcome.
A defensible basis for a buy, invest, renegotiate, investigate further, or stop decision.
Typical engagements begin at $50,000. The statement of work defines access, test conditions, deliverables, acceptance criteria, and the decision deadline. Missing access and unresolved claims are recorded explicitly.
Assessment scope and evaluator independence are agreed before work begins. Findings document confidence, limitations, and residual risks.
Review the assurance method ↗