Where the 97% figure comes from#

The 97% figure comes from a January 2026 WAV Group and Delta Media survey on AI usage among US Realtors®. It measures the percentage of agents who report using at least one AI application in their daily activity. The number jumped from 80% in 2024 to 97% in early 2026—an unprecedented adoption curve in real estate.

But that headline number hides depth. According to NAR’s 2026 technology survey, only 23% of agents have rebuilt at least one core workflow around AI (listing creation, market analysis, lead nurturing). The other 74% paste a prompt occasionally and move on. The proptech offer exists—Zillow, Compass IQ, HouseCanary, Redfin, Reonomy. The depth of adoption doesn’t. To understand exactly what generative AI applied to real estate covers and why it changes the game, we’ve published a guide that lays the foundations.

This depth gap isn’t explained by a lack of tools. It’s explained by human and organizational blockers most brokerages haven’t yet removed.

The 4 blockers stopping deep adoption#

Photo: Yan Krukau · Pexels

Blocker #1: “I don’t have time”#

It’s the most cited blocker. Real estate agents work under constant pressure: showings, valuations, follow-ups, paperwork. Adding a new tool feels impossible when the day is already full.

The paradox: AI is precisely the answer to that time crunch. An agent spending 40 minutes per showing report can cut it to 2 minutes with the right workflow. But to gain time with AI, you first have to invest a few hours setting it up. That upfront investment is what blocks most brokerages.

Blocker #2: “It still works without it”#

Plenty of brokerages keep signing listing agreements and closing deals without going deep on AI. “Why change something that works?” The problem: today’s “it works” is becoming tomorrow’s “it’s not enough.”

Buyers’ expectations have shifted. Competitors who go deep on AI produce more attractive listings, respond faster, present better. The drop-off is gradual and invisible: you don’t lose a listing because you don’t use AI—you lose it because the competitor who does showed up with a more convincing pitch.

Blocker #3: “It’s too technical for me”#

AI’s image is still associated with code, algorithms, and technical complexity. That perception is outdated. In 2026, using ChatGPT or Claude is as simple as sending a text message. Nothing to install, nothing to code, nothing to configure—at least for basic use.

The learning curve is short: most agents see a gain on their first use. And advanced features (Claude Projects, custom GPTs) only take a few hours to learn.

Blocker #4: “My market is unique”#

Every brokerage owner is convinced their market is unique and generic solutions don’t apply. It’s partly true—and that’s exactly why modern AI tools allow personalization. A Claude Project configured with MLS data from Austin condo loft conversions won’t produce the same results as one configured for a rural Texas market. AI adapts to the context you give it.

What deep-adopters actually report#

Brokerages that have gone deep report convergent results. The most frequently cited gains concern three areas.

The first is writing time. Listings, showing notes, emails, and presentations that used to take hours now ship in minutes. That freed-up time gets reallocated to prospecting, appointments, and client relationships—the activities that directly generate revenue. The fastest ROI is on writing: here’s our method for writing punchy listings with AI, prompt by prompt.

The second is perceived quality from clients. Better-written listings, retouched photos, structured CMA decks: the level of service visible to seller or buyer rises, which builds trust and makes the listing agreement easier.

The third is managing the brokerage. Brokers using AI to summarize showing feedback, analyze portfolio trends, or prep team meetings have a clearer, faster view of their business.

The gap closes in a few weeks. The competitive advantage starts building day one.

Photo: Kindel Media · Pexels

The 5 steps to take action#

Step 1: try a simple use case this week#

Open ChatGPT or Claude, and write a listing with it. Compare the result with what you would have written yourself. This first experience is usually enough to convince you to continue.

Step 2: identify the 3 most time-consuming tasks at your brokerage#

For each agent: which tasks take the most time and bring the least value? Writing, photo retouching, document reformatting, email follow-ups are almost always on the list. These are the first candidates for automation.

Step 3: configure a permanent assistant#

Create a custom GPT in ChatGPT or a Project in Claude with your brokerage’s templates, tone, and data. This step takes one to two hours and durably transforms the entire team’s productivity.

Step 4: train the entire team#

Individual gains become collective gains when the whole team uses the same tools with the same conventions. One hour of training is usually enough to make a team autonomous.

Step 5: measure and iterate#

Track simple metrics: time saved per week, number of listings produced, lead response time, seller feedback quality. Before each seller appointment, it’s also worth prepping your next seller appointment with AI to walk in with a solid pitch. This data lets you adjust and convince the last holdouts on the team.

Where to start concretely#

The 74% of US agents stuck on the surface aren’t there by choice. They’re stuck because they underestimate what’s possible and overestimate the difficulty. The depth gap closes in a few weeks. The competitive advantage starts building day one.

Why most agents stay on the surface#

The headline “97% of US agents claim to use AI” hides a subtler reality: most of them tried it once, got a generic output, decided it didn’t work for their market, and concluded too quickly that “AI isn’t there yet for real estate.” That conclusion is wrong, but it’s shared by 9 out of 10 agents based on field interviews. Understanding the real blockers lets you bypass them in weeks, while others spend two years going in circles.

Three structural blockers come up consistently:

• The “magic ChatGPT” myth: you type a vague question, get a banal response, decide the tool is useless. The reality is that a poorly framed prompt produces a mediocre result, the same way a fuzzy instruction to a new hire does. Prompt discipline is the #1 skill to develop.
• No context loaded: without feeding ChatGPT or Claude your pricing, listing templates, and tone of voice, the AI produces generic copy. One hour invested in building a Claude Project that knows your brokerage multiplies the quality of every output tenfold.
• Solo testing: an agent who tries alone, with no follow-up, no sharing, no measurement, gives up in 6 weeks. A team that trains together, shares prompts, and tracks results weekly doubles productivity in 3 months.

Brokerages that cleared these three blockers aren’t more technically gifted — they simply understood that AI is a team project, not a personal gadget. Once that shift happens, scaling takes weeks, not years.

Practical implication for any agent reading this in 2026: don’t measure your AI maturity by how many tools you’ve tried. Measure it by how many of your weekly tasks now run partly on AI, how many prompts your team shares as a common asset, and how comfortable you are turning off a tool that no longer serves you to swap it for a better one. Brokerages that treat AI as a portable skill rather than a fixed software stack are the ones that compound their lead through 2027 and beyond.