Technical due diligence
Technical DD is an objective assessment of the technical risk, technical advantage, and engineering organization behind an investment decision. With AI, investment teams can do much of it themselves. This page sets out how to run it yourself, where AI falls short, and when to bring in an expert.
What a technical DD evaluates
The purpose of a technical DD is to gather what the investment decision needs. Think of it in three areas.
Technical risk
- Technical debt and design distortions
- Security and OSS licensing
- Scalability limits
Technical advantage
- Differentiators and how hard they are to copy
- Depth of data and integrations
- Quality and unit cost of AI products
Engineering organization
- Key-person dependency
- Development process and quality gates
- How technical decisions get made
What you can do yourselves
Before outsourcing, try it yourselves first. Each step lists the articles to use and where AI helps. For the full procedure, including materials, sample questions, evidence handling, and confidentiality, see running technical DD yourselves with AI.
-
STEP 1
Understand the full picture
Learn the seven evaluation axes and how their weight shifts by investment stage.
Have an AI read the target's materials and list the issues under each axis.
-
STEP 2
Line up what to check
Use a checklist to separate what the materials can confirm from what needs an interview.
Have an AI compare the checklist against the materials and list what is still unconfirmed.
-
STEP 3
Confirm through interviews
Even without reading the source code, well-designed questions reveal much of a team's technical strength and organization.
Have an AI draft follow-up questions from contradictions or vague passages in the materials.
-
STEP 4
Turn findings into a decision
Sort findings into walk away, proceed with conditions, or acceptable, and feed them into the investment decision.
Let an AI draft the classification; people make the final call.
The limits of AI-based assessment
AI is good at surfacing issues, but a technical DD cannot be completed by AI alone. Here is where each party fits, and the conditions under which human verification is essential.
| First pass by AI | Reading and summarizing materials, listing issues per axis, mapping claims to evidence, drafting follow-up questions |
|---|---|
| Checks by your team | Interviews, checking screens, logs, and records, judging how reliable the evidence is, classifying findings, and making the investment decision |
| Verification by an expert | Hands-on review of source code, infrastructure, and security; verifying an AI product's models, data, and evaluation setup; assessment as a third party |
- You are relying only on submitted materials
- The target chose those materials. Verify through interviews with the engineers on the ground
- You have not seen the source code or production environment
- AI cannot detect gaps between design documents and the implementation. Look at the code and configuration
- You are assessing how operations really run
- Incident response and deployment practice are rarely written down. Ask specific questions and check logs
- You are assessing people and the organization
- Key-person dependency and attrition risk do not show up in documents. Talk to several people individually
- You are assessing an AI product's performance claims
- Demos and self-reported numbers may not be reproducible. Ask for the evaluation data and procedure
- You are making the final investment decision
- Use AI to surface issues; people own the decision and the accountability
When to bring in an expert
For deals like the following, do not rely on an in-house assessment alone; consider an expert assessment.
- The investment committee or co-investors require an independent, third-party assessment
- The source code, infrastructure, or security needs a specialist review
- An AI product has to be verified down to its models, data, and evaluation setup
- Evaluating in-house would create a conflict of interest, or nobody in-house can do it
Working with Tied
We take on assessments only for deals that need specialist judgment or a third-party review. Scope, team, and approach are agreed case by case.
Get in touchFrequently asked questions
Can we run a technical DD entirely on our own?
Much of it, yes. With a checklist and an AI, investment teams can organize the evaluation axes, review materials, run interviews, and classify findings themselves. Consider an expert when you need a specialist review of source code or production systems, or an independent third-party assessment.
How much of a technical DD can AI handle?
AI helps a great deal with reading materials and surfacing issues, checking them against a checklist, drafting follow-up interview questions, and drafting the classification of findings. It cannot judge what the materials leave out, how operations and the organization really work, or whether an AI product's performance is reproducible.
When should we not take an AI's assessment at face value?
When you are relying only on submitted materials, when you have not seen the source code or production environment, when you are assessing people or the organization, and when you are assessing an AI product's performance claims. These need human verification. Do not delegate the final investment decision to AI.
When should we bring in an expert?
When an independent third-party assessment is required, when the source code, infrastructure, or security needs a specialist review, when an AI product must be verified down to its models, data, and evaluation setup, or when evaluating in-house would create a conflict of interest.
Is technical DD the same service as technical advisory?
No. Technical DD is a self-contained assessment for an investment decision. If you need technical advisory after the investment, we offer it separately as our technical advisory service for startups.
Post-investment technical advisory is a separate service. See technical advisory.