I’m letting three different AIs build a real-money investment portfolio — using money I make selling things from my attic.
This is AI VC Round 3.
The three new picks are:
TransMedics — TMDX
Cellebrite — CLBT
Intuitive Machines — LUNR
That takes the portfolio to 9 companies and £90 invested.
The existing portfolio is:
Rocket Lab — RKLB
Astera Labs — ALAB
AST SpaceMobile — ASTS
Monday.com — MNDY
Recursion Pharmaceuticals — RXRX
CRISPR Therapeutics — CRSP
For the first six rounds, every funded company has to be a new company that hasn’t previously received money from any AI in the experiment.
If an AI’s true highest-conviction choice is something already in the portfolio, I still record that choice as a signal — but it then has to give me its strongest eligible new pick.
If two AIs independently choose the same new company in the same round, that overlap is also recorded as a signal, while keeping the funded portfolio diversified with new picks.
The Round 3 prompt
You are participating in the AI VC experiment.
Your job is to choose one publicly traded company for a real-money, long-term investment portfolio.
The investment amount for your selection is £10.
The intended holding period is approximately 10 years, to around 2036.
This is not a short-term trading exercise. I am looking for companies with the potential to produce exceptional long-term returns, while accepting that some picks may fail completely.
The company must be publicly traded and realistically investable through Trading 212 Europe.
The following companies have already received funding and are therefore not eligible for another funded position during the first six rounds:
Rocket Lab — RKLB
Astera Labs — ALAB
AST SpaceMobile — ASTS
Monday.com — MNDY
Recursion Pharmaceuticals — RXRX
CRISPR Therapeutics — CRSPIf your true highest-conviction choice is one of those previously funded companies, state that first. I will record it as a duplicate conviction signal.
You must then choose your strongest eligible new company for this round.
If another AI independently chooses the same eligible company during this round, that overlap will also be recorded as a consensus signal. The portfolio should still end the round with new funded companies rather than funding the same position twice.
Research using current information available at the time of the decision.
Please provide:
Research date
Highest-conviction duplicate, if applicable
Company name, ticker and exchange
Approximate market capitalisation
Investment thesis
Why now?
10-year / outlier scenario
Estimated probability of that scenario
Strongest bear case
What would kill the thesis?
Financial resilience
Competitive position
Trading 212 Europe availability
Previous-position eligibility checkFinish with:
INVEST £10: [COMPANY] ([TICKER])
And give the single most important reason for the decision.
The point of this experiment isn’t to prove that AI can beat professional investors.
I’m interested in what happens when different AIs are given the same rules, the same horizon, real money and enough time for the decisions to play out.
And because every pound invested is coming from things I’ve sold from my attic, the experiment effectively has its own little funding engine.
Nine companies down. Plenty of time to see whether any of them turn into something interesting.
Need help thinking something through?
Book a 60-minute Life Strategy Consultation:
https://cal.com/lifestrategyconsultant/
🌐 Life Strategy Consultant
Disclaimer:
This is a personal real-money experiment for educational and entertainment purposes. It is not financial or investment advice, and nothing here is a recommendation to buy, sell or hold any security. Investing involves risk, including the possible loss of capital. Always do your own research and make decisions appropriate to your own circumstances.
