// Practice areas · Commercial real estate
AI for commercial real estate
A deal practice runs on paperwork that never closes a deal: listing flyers, lease abstracts, owner reports, market notes. That layer is where AI earns its keep; the relationships, the tours, and the negotiation stay exactly where they are.
Taking first engagements: nothing is for sale until it has shipped for real users.
// The screen
Every idea runs four tests, in order
Most "AI for your industry" lists skip the part where ideas die. Mine doesn't: an idea earns a build only if it survives all four, and the ones that fail are listed below with the test that killed them.
1 · Allowed?
Licenses, data terms, regulation, confidentiality.
2 · Possible?
Can today's tools really do it, on your real data?
3 · Worth it?
Hours or dollars back, against what it costs to build.
4 · Best use?
Of the survivors, one gets built first.
// The opportunity menu
Where I'd look first in a deal practice
Time figures are industry estimates, not measurements of your business; a working session replaces them with yours. Badge key: Build first = start here · Analyzed = screened, viable · Deferred = viable, waiting on a prerequisite · Cut = failed a test, with the reason.
01
Listing & marketing collateral
Property facts and photos in; a finished flyer, listing copy, and outreach drafts out, with the license block stamped on every sheet. Hand-built, this runs 2–4 hours per listing (industry estimate).
02
Lease & CC&R abstraction
Long leases and CC&Rs distilled to the deal points that matter (term, escalations, options, use restrictions), with potential exclusive-use conflicts flagged for review. There's a reason an entire abstraction industry bills by the hour for this: done by hand, one document takes most of a working day (industry estimate).
03
Market notes & newsletters
A recurring market note in your voice, for your farm area, the kind of steady presence that compounds into referrals.
04
Landlord activity reporting
The recurring owner report (tours, prospects, comps) assembled from the week's activity instead of rebuilt from scratch each month.
05
Tenant–site matching
Rank which retailers fit a vacancy by published site criteria. Real, but no such criteria dataset exists off the shelf; assembling one would be its own project.
06
CRM & pipeline automation
Usually the first thing pitched, and usually wrong: the CRM belongs to the brokerage, so an outside tool inside it is their IT department's call, not a consultant's to make.
07
Valuation & BOV drafting
Comps platforms license their data with terms that bar feeding it to outside tools. I don't design around another company's contract.
// What I'd leave alone
- The relationship. Prospecting calls, tours, the negotiation itself. Being the broker people trust is the product; automating it away is malpractice.
- Fiduciary judgment. Which offer to recommend, what to counsel an owner: the licensed part of the job stays human.
- Licensed data. If a data provider's terms say no extraction, that's the end of the conversation.
- Legal opinions. An abstract summarizes; it never advises. Conflicts get flagged for counsel, not resolved by a model.
A tool that knows its limits is the only kind worth installing.
Accepting clients. A working session is how we find out what your version of this list looks like: your workflow, your numbers, your priority order. Early engagements are how a service earns its place on this menu, so if the fit is right, let's talk.
Book a working session