We spent a year automating our acquisitions operation.
Now we are doing the same thing to property management. And it is harder.
The acquisitions machine made sense to build first. Deals come in, you underwrite them, you close the good ones. The workflow has a shape. Once you have mapped it, you can start automating it piece by piece. We've closed 6 deals in the last year with 2 more under contract with a small team, using AI to help with everything from underwriting to drafting investor decks.
Then we turned around and looked at property management.
PM does not have a clean shape. It is a living system. Tenant calls. Work orders. Lease renewals. CAM reconciliations. Vendor bids. Rent collections. Compliance deadlines. Each one is its own workflow, and each one shifts depending on the asset class, the market, the tenant, and the history of the relationship.
My sister Kelly Chang Levine runs our PM operation. The first thing she said when we started this process: you have to understand every single task before you can touch any of it. So that is what we did. We mapped every workflow. Every exception. Every place where a person was doing something that looked simple from the outside but was carrying twenty years of institutional knowledge inside it.
Off-the-shelf software is built for the average real estate company. Which means it is built for no real estate company in particular. Every real estate company is structurally different in ways that go all the way down.
When you buy a platform, you bend your operations to fit the software.
When you build your own skills on Claude Code, you build the software to fit your operations.
We have a skill for tracking critical lease dates across our industrial portfolio. A skill for routing work orders that knows our vendors, their scope, and when to escalate to Kelly. A skill for rent collection follow-up.
None of these exist in any vendor's catalog. They cannot. They are ours.
The constraint is not the technology. The constraint is knowing your own operation well enough to tell it what to do.
That is the real work. And it turns out to be more valuable than any platform demo we have ever sat through.
Are you building your own tools, or waiting for someone to hand you the right one?
The Hard Part of AI Is Knowing How Your Business Actually Works
