From Hype to Hard Numbers: My First Real Year With AI’s in Property Development
Using AI to collapse uncertainty in design, permitting, budgeting, and execution
At the end of 2024, I set a few days aside to start experimenting with some of the free AI tools that were being crazy‑hyped to see if and how I could incorporate them into my workflow. I was not impressed. I wrote about my frustrations in Testing AI’s Limits: A Real Estate Developer’s Experiment, where the platforms I used failed to create basic arithmetic representations of simple real‑world problems. I am sure you probably remember these types of word problems from your 4th and 5th grade days. The questions usually involved other kids as characters, some number of fruit, or some trains traveling at some speed, all of which you needed to decode to create a basic arithmetic calculation. Moreover, at the end of ’24, free AI platforms could not arbitrarily retrieve data from the web or access files on your local drive. Needless to say, in 2024 things were more hype than reality. But it is now 2026 and oh how things have changed.
Over the last year and a half, the forces driving adoption of AI have been irresistible. Product enhancements enabling active retrieval of online information, composition, and its presentation have dramatically increased the usefulness of these tools for property developers. A second area of enhancements has also made a profound difference in the usefulness of these tools, namely the ability to access local archives of information. This functionality means that the individual context of each deal and the uniqueness of its structure and location can all be considered when working to solve the hard problems. As a developer working on an individual property in an individual town, most of the critical project context will come from the work done by architects, engineers, and lawyers. In other words, your contracts and your designs.
In the last few weeks, I have come to realize that the improvements in these AI products are really beginning to add value to the work I do. There has been a great deal of chatter in real estate media and online communities about the use of AI to relieve the drudgery of repetitive tasks, specifically tenant communications and administrative tasks. Since I do not manage properties on behalf of others, these are not tasks I am trying to avoid. I want to know my tenants. I want to know what is going on with my building. So for me, adding a layer that dissociates me from my business seems unwise. I keep my properties at the highest level of maintenance and work hard to select reliable people to occupy my units. If things don’t break and neighbors act decent, you get very few calls, emails, or texts from tenants. For those times when tenants do reach out, a simple appointment with an appliance repair vendor, exterminator, or plumber will be enough to resolve the issue. You don’t need to triage problems if you invest to prevent a majority of them in the first place. That is why, for me, the value in AI tools is more closely linked to property development strategy and financing than tenant request handling and scheduling.
As someone that started their career in municipal government and later worked in market research and business strategy, I think I am exceptionally good at researching government and policy‑related stuff. And that is exactly where I found AI tools to create the most leverage for me. Between the state and federal historic restoration incentives, state and federal energy efficiency incentives, state and local industrial development initiatives, federal economic development incentives, and public/private initiatives, there are so many opportunities for benefits to the project that it is almost too hard to wrap your mind around if you have not been in commercial property development. While most of this is somewhat new to me, my real estate and municipal government experience gives me just enough knowledge to make me dangerous. Why? Because it used to be required that you were smart enough to ask the right questions AND find the right answers. With AI, you only need to be half as good, you only need to know how to ask the right questions. So, I asked a whole lot of questions about any type of program I could find, and used AI to find some more.
We knew our current project in Troy was in the historic district before we even owned it. What we did not know was if the building was qualified, the magnitude of incentives, and the cost of compliance. Our building, a brick Italianate from the 1830s with a 1900s makeover in the downtown Troy historic district, opened the door to the Federal Historic Rehabilitation Tax Credit. The program offers a 20% credit on qualified rehab costs for income‑producing buildings listed on (or eligible for) the National Register or located in a historic district. Since 80 4th Street is a contributing structure in Troy’s district and our renovation is substantial, we initially qualified. New York State stacks another credit on top, up to 30% for small projects. But certified historic restoration comes with a steep premium: SHPO‑approved materials, preservation architects, historically accurate windows, and NPS‑compliant methods can push a $1.5 million project toward $3 million. Without large capital gains to offset, the economics flipped. The credits were powerful, but the cost structure only made sense for investors playing a different tax game.
We also discovered that the building was in a Federal Opportunity Zone, which was renewed shortly after we acquired the property. OZ rules allow investors to roll capital gains into a project, defer taxes, and pay no capital gains tax on appreciation after a 10‑year hold. For someone investing their capital gains, that could mean six‑figure tax savings if the property appreciates. To qualify, we structured the project as a Qualified Opportunity Fund and confirmed our renovation easily met the “substantial improvement” test. OZ status didn’t give us free money, but it unlocked cheaper equity, an advantage we wouldn’t have spotted without the help of AI.
Energy incentive questions led us to find another layer of programs. The federal 179D deduction could save us several thousand dollars for energy‑efficient upgrades, while the 45L credit offers $2,500–$5,000 per apartment for high‑performance units, potentially $30,000–$60,000 across our 12 planned units. NYSERDA programs and utility rebates for heat pumps, insulation, and solar could add tens of thousands more. AI helped us identify the exact efficiency targets, certifications, and paperwork required, turning what used to be a maze of obscure programs into a clear checklist.
On the local side, the Troy Industrial Development Authority became a major win. After reviewing past deals and submitting an application, we are going to pursue a 10‑year PILOT agreement that will save on property taxes, provide a sales‑tax exemption on materials for the building project, and a $10,000 mortgage‑recording‑tax waiver. These savings directly improve project viability and allow us to reinvest in better finishes and systems. AI helped us parse through the program guidelines with ease to understand our qualification and potential benefit.
Leveraging AI made it possible for us to discover incentive programs we did not know existed, helped us avoid wasting time on programs we would not be eligible for, and helped us to quickly quantify and prioritize all the others. When you start working with these tools, the first shift is learning to treat AI like a junior analyst with infinite stamina. It won’t guess what you want, and it won’t magically intuit the structure of a problem. But if you give it clear constraints, defined outputs, and a well‑framed question, it will return leverage that compounds. The second shift is realizing that none of this works without a local archive. Your contracts, drawings, pro formas, emails, and municipal correspondence are the real intellectual capital of a deal, and the moment you centralize them, cleanly, consistently, and in a way that’s queryable, you unlock a level of clarity that simply didn’t exist before. If the AI knows enough about your project it can tell you if you will meet the qualification criteria. From there, the discipline becomes using AI as a second brain. Run your numbers manually, then run them again through the model. When the two disagree, don’t panic; investigate. The gap is where risk hides, and closing that gap is where judgment sharpens. AI is often right, but sometimes wrong. Don’t trust, just verify.
As you move deeper into a project, you’ll notice that AI is most valuable in the fog‑of‑war phase, the stretch where you’re still decoding the architect’s drawing, comparing contractor bids, or trying to understand the real constraints buried inside a code section. The faster you collapse uncertainty, the faster you can move, and these tools are built for exactly that kind of compression. As a decision maker, I can now gather far more data and throw my research tentacles much farther than Google’s basic search engine ever let me do. Simply put, I can only take advantage of programs and incentives if I know they exist. If you can scale up your ability to gather and synthesize information every decision you make should yield better outcomes.
In the end, the developers who thrive in this new era won’t be the ones who memorize the most rules or grind the longest hours. They’ll be the ones who learn to direct intelligence, human and AI, with precision. AI doesn’t replace judgment; it clears the noise that used to cloud it. When you pair your lived experience with a tool that can retrieve, summarize, and structure information at scale, you stop operating as a single mind and start operating as a system. That is the real leverage. And once you feel it, you won’t go back.
Ready for more insights from the field? I’ve made the expensive mistakes, so you don’t have to. Subscribe to get weekly strategies, real numbers, and unfiltered lessons from 25+ years of actual deals. No theory, no fluff, just lessons learned and what actually works when your money is on the line.