AI & Tech

The AI Opt-Out Is Over for Brokerages. The CRM Is the Real Battleground

Almost every brokerage now uses AI. The ones that pull ahead will be the ones that wire it into the CRM, with guardrails, instead of bolting on another tool.

Eduardo Diaz · 2026-08-28 · 5 min read

real estate CRM AI adoption brokerage operations

Two years ago, about one in ten brokerages told Delta Media Group they had no plans to adopt AI. This year it's fewer than one in fifty. Delta's three-year analysis, released August 5, puts the share of brokerages with no AI plans at 1.9% in 2026, down from 10.6% in 2024. The holdouts are basically gone, so the interesting question has changed. It's no longer whether your brokerage uses AI. It's where.

What the numbers actually say

Delta Media runs an annual Real Estate AI & Leadership Survey of more than 100 brokerage leaders. According to HousingWire's coverage of the 2026 results, those leaders represent firms responsible for more than two-thirds of U.S. real estate transactions in the prior year. That isn't a niche sample.

Here's what stood out to me:

Delta's CEO Michael Minard put it plainly: "Every year we run this survey, the story is less about who is trying AI and more about who still has not."

Adoption is solved. Usefulness isn't.

Look at what agents mostly use AI for. In the 2026 survey, roughly 82% of agents used it to write listing descriptions, and 74% used it for blogs, social posts and email campaigns. Florida Realtors reported on August 25 that content such as marketing materials is still the top primary use.

That's fine. It saves time. But it's also the shallow end of the pool.

A nicer listing description doesn't fix the lead that came in at 9:40 on a Saturday night and never got a reply. It doesn't remind an agent that a buyer's pre-approval expires next week. It doesn't notice that a past client just started browsing again. Those moments live in the CRM, and that's where most of a brokerage's revenue is either won or quietly lost.

Every brokerage owner I talk to says some version of the same thing. Their agents use ChatGPT. Their CRM still runs the same way it did five years ago. The AI sits in one browser tab and the business sits in another.

Why the CRM is where this gets decided

The leaders in Delta's survey seem to see it too. HousingWire reported that brokerages plan to expand AI into workflow automation, marketing automation and agentic AI that can carry out multi-step tasks. WAV Group's write-up lists CRM systems, administrative and back-office tasks, and recruiting and training among the planned areas. Respondents also described consolidating around platforms that bundle AI, automation and CRM instead of buying standalone tools.

That shift matters because your CRM is your system of record. It knows who the lead is, which agent owns them, what was said last, and what should happen next. An AI tool that can't read or write to that record is guessing. An AI that works inside it can act.

In practice, "AI in the CRM" should mean things like:

None of that is exotic. It's just AI doing the boring, high-stakes work in the place where the work already happens.

The guardrails question is real

Here's the part I don't want anyone to skip. Leaders are not naive about the risks. WAV Group reported that 49% of brokerage leaders rate their concern about AI guardrails between 7 and 10, and that the top barriers include data privacy risks, compliance requirements and integration challenges. Florida Realtors noted that concern about safeguards averaged 6 out of 10, unchanged from the previous two surveys, with data and security risks ranked highest.

So adoption went up and the worry didn't go down. I think that's healthy.

My background is in healthcare, where you don't get to be casual about who sees what data or what a system says on your behalf. Real estate has its own version of this: fair housing, licensing rules, consent for texting, and client financial details sitting in a contact record. When AI lives inside the CRM, you can actually control those things. You decide what the model can see, what it's allowed to say, when it escalates to a person, and what gets logged. When AI lives in a dozen disconnected tools, you mostly hope.

Bolting on more tools makes the guardrail problem worse, not better. Wiring AI into one system of record is how you make it manageable.

Questions to ask before your next AI purchase

If you run a brokerage and you're deciding what to do next, these are the questions I'd put on the table:

  1. Does it write to our CRM, or just read from it? If the output ends up in a chat window, someone still has to copy it over.
  2. What happens to a lead in the first five minutes? Walk through a real inbound text or form fill, step by step, and see where it stalls.
  3. Who owns the follow-up when the agent is busy? If the honest answer is "nobody," that's your first automation.
  4. What data can the AI see, and what can it say? Get this in writing. It should be configurable, not a black box.
  5. When does a human take over? Good systems hand off cleanly, with full context, before a conversation goes sideways.
  6. Can we measure it? Response time, contact capture and missed follow-ups are simple numbers. If a tool can't move them, it's not doing much.

What this looks like when it's built

We built an AI-powered SMS CRM for a real-estate brokerage using Google Gemini 2.5. Leads are captured in about 30 seconds, contact capture is close to 100%, and the brokerage now has zero missed follow-ups. You can read the details in our CRM case study. The point isn't the model. It's that the AI does its work inside the system the team already relies on, so every conversation ends up as a record someone can act on.

If you're weighing where AI should actually sit in your brokerage, I'm happy to talk it through. You can see more about how we approach this on our real estate page, and I'd start with whatever is costing you leads today.

Sources

Keep reading

← Back to Insights