First investments are to be made at the beginning of 2026. Check back Q2 2026 for initial results.
There is an opportunity for a coordinated group of independent agents to reproduce the due dilligence of a large value investment firm, to arrive at high confidence buy-and-hold stock recommendations over long time horizons.
There are multiple areas where code agents can reproduce the work of actively managed funds. By dramatically lowering the expense ratio the models can be even more conservative with their picks than humans while still maintaining an attractive alpha.
All buy and sell decisions will be made by code agents
High level strategy will also be made by code agents
Human guidance will be in the form of writing new code only
No margin and no fast trades
The platform continuously ingests a comprehensive stream of global market data, corporate filings, and financial news into a unified analytical engine. This data is then analyzed in parallel by a suite of distinct AI agents, each employing a different investment philosophy, from deep value to reinforcement learning, to identify unique market inefficiencies. Finally, a master AI synthesizes these quantitative findings with deep qualitative reasoning to construct a concentrated, low-turnover portfolio of high-conviction ideas designed for long-term capital appreciation.
The system is built on a serverless Google Cloud architecture, where an automated data pipeline leverages Cloud Functions and Cloud Scheduler to populate a central BigQuery data warehouse. From there, containerized AI agents on Cloud Run perform quantitative analysis, with Vertex AI providing a final qualitative synthesis to guide the fund's long-term portfolio decisions.
The fund will not be public. It will follow the attitude of Renaissance Technologies in that it will serve primarily as my own investment vehicle. If the approach is successful and I have confidence in the quality of the systems I may invite friends and family to invest.
I probably will not invest more than $40k or so of my money into the fund in the first year. This is not meant to be a get-rich-quick-scheme. Over time I will move more of my savings over, and I may ultimately move my entire savings over.
I am growing tired of consumer technology and user interfaces. I want to spend more time on hard technology.
I think value investing is one of the least cynical applications of AI if it means in practice separating the signal from the noise (in a very noisy society!!). Doing this work sounds fun.