Most CRE firms have an AI pilot running somewhere right now, and most of those pilots are not doing much. Why?

I build CRE intelligence tools for a living. My engineering team and I ship features, watch how they get used, and go back to fix what is not working. After doing this for a while, I have a pretty clear picture of why the gap between piloting AI and actually getting value from it is so wide, and the answer is not the AI technology itself. The technology is ready. The problem is almost everything else around it.

The Features That Stick Are Not the Ones That Look Coolest

When we ship something new, we track whether people come back and use it again. The ones that stick are almost always the features that save time on something the user was already doing every day, not a new workflow or a new capability they have to learn from scratch, but something familiar that now works faster and without the frustrating parts.

The features that do not stick follow a pattern too. They look impressive in a demo and surface data in a visually compelling way, but after the demo, the user asks "so what?" and cannot come up with a good answer. The tool organized information in a sophisticated format, but did not unlock something they didn’t already know.

If a feature does not directly help someone close a deal, build a prospecting list, or answer a client question faster, it is going to go unused in commercial real estate workflows. That is just the reality of how commercial real estate professionals work, because they’re in search of tools that allow them to build better relationships and create a better flywheel of business.

AI Is Only as Good as the Data Behind It

When we ship something new, we track whether people come back and use it again. The ones that stick are almost always the features that save time on something the user was already doing every day, not a new workflow or a new capability they have to learn from scratch, but something familiar that now works faster and without the frustrating parts.

The features that do not stick follow a pattern too. They look impressive in a demo and surface data in a visually compelling way, but after the demo, the user asks "so what?" and cannot come up with a good answer. The tool organized information in a sophisticated format, but did not unlock something they didn’t already know.

If a feature does not directly help someone close a deal, build a prospecting list, or answer a client question faster, it is going to go unused in commercial real estate workflows. That is just the reality of how commercial real estate professionals work, because they’re in search of tools that allow them to build better relationships and create a better flywheel of business.

The "Before State" We Keep Solving For

Before Crexi Intelligence, the workflow for building a prospecting list or running due diligence on a portfolio looked roughly the same for almost everyone we spoke with. Users had five browser tabs open, pulling ownership records from one source, cross-referencing property details from another, manually validating addresses, copying everything into a spreadsheet, and starting over when the data did not match.

We built Crexi’s Data Enrichment tool specifically to compress that workflow. Users upload a list of addresses and receive enriched property data and owner contact information back within minutes. What used to take hours of manual lookups now happens in a single session on one platform, and for a broker building a cold outreach list or an investor screening a large portfolio, those recovered hours add up quickly.

The bigger problem we were actually solving, though, was not just time. It was the cost of constant context-switching. Every time a user leaves a platform mid-task to find information somewhere else, they break focus and introduce the possibility of error. When the information they need is already on the platform where they are working, they stay focused on the work that requires actual judgment: structuring a deal, advising a client, or reading a market. That compounding effect across a full session is hard to quantify, but very easy to feel.

The Real Reason Adoption Stalls

When I try to explain why so many CRE firms are not getting results from AI, I keep coming back to the same answer: AI education.

The CRE industry has asked a similar question for years, which is why more people do not invest in commercial real estate. The most common answer, beyond the capital requirements, is simply that people do not know how. When something feels unfamiliar, people move slowly and miss the compounding benefit of acting faster.

CRE AI adoption is following the same pattern. The technology is not the barrier, and the data infrastructure, at least on platforms that have invested in it, is not the barrier either. The barrier is that most teams have never had someone sit down with them and show them specifically where these tools help, where they do not, and how they fit into real CRE workflows.

If you are leading a CRE firm and have not made a deliberate effort to educate your team on how to use these tools well, your pilot is not going to go anywhere, not because the tool failed, but because nobody gave your team a clear mental model for when to reach for it.

What an Actual Workflow Change Looks Like

There is a meaningful difference between AI that makes a task incrementally faster and AI that changes how the task gets done entirely. We see the first one constantly and the second one far less often, though the second is where the bigger value is.

The clearest example I can point to right now is a tool we are about to release: an AI-powered zoning analysis feature inside Crexi. Today, determining what can be built or operated on a property requires navigating county websites and interpreting hundreds of PLU codes, a process that is slow, inconsistent, and often leads most practitioners to either delay or hand the work off entirely.

With the new tool, a user simply asks the question in plain language, and Crexi returns the answer directly. That is not just faster zoning research. It is a fundamentally different relationship with zoning research, where a task that used to require specialized expertise or outside help becomes something anyone on a team can complete in seconds. That is the version of AI adoption in CRE worth pursuing.

Where We Are Headed

My team and Crexi at large is currently focused on a specific question: what are users leaving our platform to do elsewhere, and how do we bring those CRE workflows into the product?

Every context switch mid-task carries a cost. Users return slightly less focused, sometimes carrying information that does not translate cleanly back into what they were already building. The best tools reduce that friction because they have absorbed enough context about how a user works to make switching less necessary. They know the markets a broker covers, the deal parameters an investor typically targets, and the data fields that matter most for a given asset class.

That is the direction we are building toward at Crexi, not a single feature but a platform that becomes more useful the more it is used, because it is accumulating context about your deals, your markets, and your working patterns over time. We are not there yet, but every product decision we are making right now is pointed in that direction.