Consult · Technical due diligence
Technical due diligence for investors.
A written report on a software or AI company, and one call to walk you through it. You get a clear view of what is real, what is claimed and what could go wrong, before you commit.
From £4,500 (light) to £9,500 (full). 5 to 10 working days.
Fixed price per deal, agreed before we start. Everything is written down, and you can send it on to your committee.
What we check
Six things, each written up with the working shown.
Architecture
How the system is put together, where it will strain as it grows, and what depends on a single person, vendor or service.
Code quality
Whether the code can be read, tested and changed safely: tests, structure, dependencies and how releases go out.
AI claims, tested
Each claim the company makes about its AI is set against evidence. Where a figure can be re-run, we re-run it. Where it cannot, we say so.
Security
Access, secrets, data handling and the obvious ways in. A focused review, not a penetration test.
Team and process
Who built what, how work is reviewed, and what happens if a key person leaves.
Risks
A ranked list of what could hurt the investment, with how likely we think it is, what it would cost to fix, and what we could not check.
Two levels
Light or full, by the size of the deal.
Light · from £4,500
Five working days. A review of what the company shares, focused on the AI claims and the top risks. Best for early rounds and smaller cheques.
Full · up to £9,500
Up to ten working days. Adds a hands-on look at the code and infrastructure and a review of security and team process. Best for larger rounds.
Timing starts when we have the access we need. One call at the end, usually 45 minutes, to go through the findings and take your questions.
Sample report outline
What lands in your inbox.
- 1. Summary and recommendation, on one page
- 2. What we were given, and what we were not
- 3. Architecture and where it will strain
- 4. Code quality and delivery practice
- 5. AI claims table: each claim, the evidence, and our finding
- 6. Security and data handling
- 7. Team, process and key-person risk
- 8. Ranked risks, with cost and likelihood
- 9. Questions to put to the founders
- 10. Appendix: method and sources
Free checklist
Forty questions to ask an AI company before you invest.
Claims and evidence, data, model and vendor dependence, testing, human oversight, security, costs, team and regulation. Use it yourself or send it to the founders. Printable, with a PDF.
How we work
Straight findings, no strings.
- Every finding says where it came from: a file, a test we ran, or something the team told us.
- Where we could not check something, the report says so instead of guessing.
- We work under a confidentiality agreement, and we hold no stake in the company or the deal.
- Jon Ross writes and signs every report. AI tools help with reading and checking, and a person reviews every line.
VOCABOTICS LTD, registered in England and Wales (no. 17412438). Cyber Essentials certified, with professional indemnity and cyber insurance.