Can software invest like Buffett?

Most market noise is exactly that. Buffett's edge was never speed, it was patience and a short list of questions asked relentlessly: does this business have a moat, is management honest and rational, and is the price below what the thing is actually worth? I am trying to teach software to ask those same questions, slowly and on purpose.

The hard part is not the finance, it is the discipline. An LLM that sounds confident is a liability in a domain that punishes hand-waving. So most of the work is making it slow down, cite the evidence behind every claim, and say I don't know when the data is thin.

What I am building

• Reads the filings. Pulls annual reports and statements and turns a decade of them into structured numbers, not vibes.
• Scores the moat. Looks for the durable advantages: pricing power, returns on capital, the things that hold up over time.
• Weighs management. Capital allocation, insider behaviour, and whether the story in the letter matches what the numbers say.
• Margin of safety. An honest range for what the business is worth, and how far below it today’s price actually sits.
• Stays disciplined. Every call carries the evidence it used, and gets graded later against what actually happened.

Follow along

It is early and opinionated, and it is where a lot of my writing about LLM discipline and evaluation is going to come from. The notes get published as I go.