Bo Eason wrote a book called There Is No Plan B for Your A-Game, and in it he says: “Tell your story.” So here is mine.
After earning my Wharton real-estate investment & analysis certificate, I was asked to evaluate the feasibility of a large new apartment community. I went to the site, took photos, visited a dozen comparable properties, studied the schools, the road noise, the special features — a full top-to-bottom read of what the finished project would be worth against current market rents, minus capital expenses, using the right cap-rate math. I pulled a favor from a top apartment broker for past-sales data, and another from a large management company to sanity-check my capex ratios against my own experience. I was told the family office looking at the deal said, “We need a guy like this” — because no stone was left unturned.
The ask that should have been impossible
Then came a bigger call: evaluate a large, capital-intensive regional casino-resort development — golf, retail, housing, and the casino itself, all one project. What would it cost to build? What would it be worth when finished? Could I have it done in two weeks? (I keep the client and the specifics confidential; the details here are illustrative.)
I thought to myself: you have to be kidding me. Yes, I have a construction-management degree from UC Berkeley, I’m a licensed general contractor, I hold a real-estate license and the Wharton analysis certificate. But I didn’t know the first thing about running a casino, or what the numbers look like on one. I sat with it for a few days. There was just no way — too much work, months to do it properly.
And then it hit me: what about AI?
I taught my AI interface the lessons from the Wharton certificate. I taught it to examine the project against the five construction-cost methods I’d been learning since college and using in the field ever since. I used AI to research other projects of the type — sale prices, square footage, the kinds of operators involved. I watched hours of background on the area, called the local broker’s office and asked for their honest read on viability, bought a basic modeling format and trained the AI on it. Then I ran it all together as one cross-referenced package. And something was off — the off-sites allowance came in too low for my taste. So I kept working the trained model until that number landed in the ballpark.
I made the deadline — and learned the real lesson
I made the two weeks. The client was floored by the output — and by the cross-checking that showed exactly where every number came from. Honestly, I was floored too. It is remarkable what is now possible with AI for rapid research and analysis. But here is the part I tell everyone: AI still hallucinates. The real trick is to already be a professional in the field, so you know what to look for, how to verify the assumptions and the outputs, and how to use AI to amplify your own skill many times over.
The work compounded
From there it kept building — a nuclear-power-adjacent venture, then a patent-monetization engagement, each one sharing sources, partnerships, and research with the last. The crossover is real: a custom news feed built for one industry sharpens the next; a grant that applies to one project surfaces for another. I work directly in the terminal now, in Python, with agents that do careful, checkable work inside clear lanes — a human deciding at every line that matters.
Why this matters to me
Watching agentic AI do genuine work — not a parlor trick, but careful, verified, human-in-the-loop work — is what convinced me this isn’t just a deal-room tool. It is a new way to amplify a lifetime of hard-won judgment, in real estate, in finance, and beyond. That is the journey I’m on, and the reason I built the rest of these pages. A new fire is here; the work is to tend it carefully with a steady hand, with proper guardrails and checks in place.
A companion piece: Taming the Fire — Keeper in the Loop →
← Back to My Agentic Journey