The Same Time Sink, Twice
My path into commercial real estate didn't start in real estate. It started in M&A, working on deals at Disney and Pixar before I moved into real estate private equity. Two very different worlds. One thing was identical in both.
The single biggest waste of time was the first-pass work. Taking an offering memorandum apart by hand. Rebuilding the model. Reconciling the numbers across the OM, the T-12, and the rent roll. Hours of it, every time, just to answer one question: is this deal even worth a second look?
That work has to happen. But the way most teams do it, a smart analyst spends the bulk of their week as a human data-transfer pipe instead of an analyst. That was the conversation I came to have with Michael Pouliot on the DealFlow Podcast, and it's where a lot of the misunderstanding about AI in this industry starts.
AI Removes the Grind, Not the Judgment
The fear you hear most is that AI replaces the junior analyst. I think that gets the role exactly backwards.
What AI removes is the manual grind, the re-keying, the formula-checking, the version reconciliation. What it leaves untouched is the judgment, and judgment is the whole job. The analyst who used to burn five hours rebuilding a model from a PDF now spends those five hours pressure-testing the assumptions inside it: Is that rent growth realistic for this submarket? Does the exit cap make sense against where rates are heading? What is the sponsor not telling us?
That's a bigger job, not a smaller one. You're asking a 25-year-old to think like a principal instead of behaving like a copy machine. The firms pulling ahead aren't cutting their analysts, they're using AI underwriting as a force multiplier and pointing all that recovered time at better questions.
Reinventing the First-Pass Review
The most concrete example we got into is the first-pass review itself. Historically, an analyst's morning is triage: an OM lands, and someone has to read it, sketch a quick model, and decide whether it clears the buy box before anyone senior spends a minute on it.
AcquiOS turns an offering memorandum into an investment-ready model in minutes, and the AcquiScore ranks every inbound deal against your firm's criteria the moment it arrives. The repetitive triage that used to eat the first half of the day happens before the analyst opens their laptop. They start from a ranked, modeled pipeline, not a blank spreadsheet, and put their energy where it counts.
It's the same logic on the diligence side. AcquiOS can verify rent comps live, with AI agents that call leasing offices to confirm asking rents and concessions, instead of an analyst chasing the same calls by hand. The output is faster and more grounded in primary sources.
The Real Risk Is the Untraceable Output
Here's the part of the conversation I care about most, and the part that doesn't get said out loud often enough. The danger in AI underwriting is not the AI. It's trusting an output you can't trace.
A confident, wrong number is the worst thing that can happen in this business. If a junior analyst takes an AI's projected NOI at face value, doesn't understand where it came from, and that figure flows into an IC deck, you've automated a mistake at speed. That's not an AI problem. That's a process problem.
It's why verifiability and citations are non-negotiable for us. Every number AcquiOS produces points back to a source, the exact page and line in the OM, T-12, or rent roll it came from. An analyst can click any figure and see where it originated. The goal isn't to get people to trust the machine. It's to make every output checkable in one click, so trust is earned line by line.
Every Deal Becomes a Data Asset
The last theme we covered is one of the most underrated. Today, most of the work a firm does on a deal it passes on simply evaporates. The model, the notes, the comps, the reasons for the pass, all of it lives in a folder no one opens again.
When the first-pass work is structured from the start, every deal you look at becomes a data asset for your firm. The 200 deals you declined last year become a searchable record of what you saw, what you assumed, and why you said no, institutional memory that compounds instead of walking out the door when an analyst leaves.
Whether you think AI in CRE is overhyped or underhyped, that's the shift worth paying attention to. The grind shrinks, the judgment grows, and the work you've already done starts working for you.
Thanks to Michael Pouliot for having me on the DealFlow Podcast. Watch the full episode here, and if you want to see what OM-to-model in minutes actually looks like on your own pipeline, book a demo.