How I Use AI Across Every Asset Class

I've written about how I use Grok and Claude to analyze a stock, read a 10-K, run the math, ask the skeptic's questions. What I haven't spelled out is that it's the exact same method for real estate, and for construction, which surprised me the first time I noticed it. The tool doesn't change. The discipline doesn't change. Only the document you're feeding it changes.

The rule that travels across every asset class

Use AI to analyze, never to predict. That's the whole rule, and it applies identically whether you're looking at a quarterly earnings filing, a rental property's rent roll, or a set of architectural plans. An AI model doesn't know what a stock will do tomorrow, doesn't know whether a specific rental market will hold up, and doesn't know whether a contractor's framing plan will pass inspection. What it can do, faster than any human, is read a document closely, compute the numbers, and hold you to whatever rules you've already decided to follow.

Stocks: the version I've already written about

For equities, this looks like feeding a filing to the model and asking for growth rate, margins, cash generation, and balance sheet health, then asking it to compute position size, expected value, and the recovery math on a potential loss before I ever place a trade. The one prompt I never use is should I buy this, because the model can't know my time horizon or my other positions. It can only do the reading and the arithmetic, which turns out to be most of the actual work.

Real estate: the same questions, a different document

Underwriting a rental property is the same exercise wearing different numbers. I'll have the model pull comparable rents for the area, run the cap rate and cash-on-cash return against the asking price, and check the DSCR math against the specific loan terms I'm considering. Then the skeptic questions: what would have to be true for this rent-growth assumption to be wrong, what happens to the return if the vacancy rate runs a full month longer than modeled, does the numbers still work if the rent-control jurisdiction turns out to cap increases more tightly than assumed. That last one matters enough that I've written a full piece on why the ordinance has to be checked before the numbers mean anything at all. AI is very good at surfacing that check, if you remember to ask it the question in the first place.

Construction: research before you call the inspector

This is the use case that surprised me most. When I've built as an owner-builder, I've used AI to review architectural plans before submitting them, describing or sharing the drawings and asking what an inspector at each stage is actually going to be checking for, what the code requirements are for my specific jurisdiction, and what questions I should have answered before I call for a foundation or framing inspection. It's genuinely useful research, the same way it's useful for a 10-K or a rent roll. It reads documents fast, and it can explain a code requirement in plain language before you're standing in front of an inspector trying to understand it for the first time.

What it cannot do is replace the inspector, or a licensed engineer's stamp where one is required. An AI model reading your foundation plan is preparation, not approval. The building department's inspector, standing at the site looking at the actual rebar and forms before the concrete truck shows up, is the one whose sign-off is real. I treat AI here exactly the way I treat it with a stock filing: it does the reading and the research so I show up to the actual decision already informed, but it never makes the decision, and it never replaces the person whose job it is to sign off.

One method, three documents

A 10-K, a rent roll, and an architectural plan look nothing alike. The method for using AI on all three is identical: feed it the actual document, ask it to compute the specific numbers or requirements that matter, then ask it the skeptic's question that would prove your assumption wrong. That's the whole system. It's not three different skills. It's one skill, and once you notice that, it's obvious why the same investor who writes about stop-loss discipline is also the one who writes about rent ordinances and foundation inspections. It was never three different newsletters.

I'm David Bellairian, founder and CEO of Glencore Associates, a Los Angeles private investment and asset management firm, and founder of AIDiscover, an AI-powered marketing and brand discovery agency. This is my own approach to using AI research tools, not investment, legal, or engineering advice. Always use a licensed inspector and, where required, a licensed engineer for construction work.