Ask AI about a stock and it’ll give you a confident, well-organized answer. That doesn’t mean it’s right.

Why I’m the one telling you this

I spent years as a Wall Street equity research analyst covering healthcare and diagnostics. Some of the most useful numbers were never public, so I found other ways to get them.

• Called clinics nationwide, asking what equipment they used and how volume had shifted, to see who was actually gaining share.

• Tracked lab workers’ social posts about new equipment, an early read on how strong a quarter would be.

• Read a serial number off a lab machine to estimate sales a company wouldn’t disclose.

I got good at telling the difference between a real answer and a confident-sounding one.

College taught the discounted cash flow model as the most rigorous way to value a company. Wall Street barely talked about DCFs. Investors cared about the multiple instead. The weakness of a DCF is that it’s only as good as the assumptions that go into it. AI has the same problem.

New models come out every few months. I test them against real research instead of headlines, and write about what actually holds up.

What free subscribers get: weekly essays on how to actually use AI for investment research. What works, what doesn’t, and why. Some weeks look at specific companies, through the same lens I used on Wall Street.

What paid subscribers get: the Tools library. A bear and bull case generator, a risk rubric, a stock classifier, and whatever I build next, pulled straight from my own process.

This is not a signals service. I will sometimes use my own positions as examples, but this is not a place to copy trades. There’s no such thing as a perfect investment, just ones that fit your risk tolerance and time horizon and goals better than another one might. If I mention a stock, I’m showing you how I thought through it, not telling you to buy it.

If you want AI to actually earn a place in your research instead of just sounding smart, you’re in the right place.