Commissioned research
Sometimes the most useful research starts with someone else's question. I take a small number of commissions. Each one is an independent, bounded piece of analysis on a specific exposure, thesis or decision involving AI.
What you bring
- An exposure you already hold: a fund position, a portfolio company, an investment whose value depends on AI economics holding up.
- A view you want tested: your own thesis, stress-tested from outside.
- A decision you're about to make: a vendor, a build, a dependency you're about to take on.
You don't need to arrive without a view. Often the most useful read is a second opinion on one you already have.
What I do
I investigate independently, drawing on the frameworks and evidentiary approach behind the published research. I don't join your team or implement anything, and I'm not paid by the vendors I analyse.
What you get
A written analysis with a defined scope:
- what I found, and the evidence for it
- the assumptions your position rests on, and how well the evidence supports them
- where it breaks: the conditions under which it fails
- what to watch, and what each signal would mean
- what the analysis doesn't cover
The work behind it
My published research sets out the frameworks and evidence behind this work. It also includes predictions made before the outcomes were known. In A Tale of Two Earnings, I published six predictions with explicit pass/fail conditions, then checked them against the companies' filings.
Read the research first. If the approach is useful, a commission applies it to your specific question.
How to start
Describe the question, the exposure or decision behind it, and any deadline. Enquiries and scoping begin in writing. I'll reply with whether it's a fit and how I'd scope it.
Commissions are confidential. Nothing is published or referenced without your agreement.
This is research, not investment or legal advice.