Will AI replace real estate agents? Not in the next few years. AI is replacing tasks, not the job: listing copy, first replies, scheduling, follow-up. In NAR's 2025 profile, 88 percent of buyers and 91 percent of sellers used an agent, a record for sellers. The agents at risk are the ones whose only value was access to information.
Key takeaways
- Buyers and sellers still hire agents: 88 percent of buyers and 91 percent of sellers used one from July 2024 to June 2025, per NAR's 2025 Profile of Home Buyers and Sellers.
- Agents are adopting AI fast. In NAR's 2026 technology report, 23 percent use AI tools daily and 55 percent say it has helped their business.
- The BLS counts 530,600 broker and agent jobs in 2025 and projects 2 percent growth through 2035, about 40,400 openings a year. Every state requires a license.
- Buyers are getting warier of AI, not less: in Cotality's 2026 survey, trust in AI tools for finding a home fell from 30 to 16 percent, and 44 percent would pay for a human to check AI decisions.
Will AI replace real estate agents?
No, not as a job, and the current data does not point that way. AI is taking over pieces of the work that are repetitive and text-based. It is not taking over the parts buyers and sellers pay for: judgment on price, negotiation, managing a deal that is falling apart, and a licensed person who is accountable for the advice.
The best evidence is what clients do. NAR's 2025 Profile of Home Buyers and Sellers, covering July 2024 to June 2025, found 88 percent of buyers used an agent and 91 percent of sellers did, tied for the highest on record. That is in the same years ChatGPT became a household name. If AI were replacing agents, the seller number would be falling, not peaking.
What AI already does in real estate
AI already writes listing descriptions, drafts emails and social posts, answers first calls and texts, schedules showings, follows up on old leads and estimates home values. Most of it is the work agents have always done between the parts clients notice. Here is the line between what AI handles today and what still needs the agent.
| Task | What AI does today | What still needs the agent |
|---|---|---|
| Listing descriptions and posts | Drafts them; 75% of agents using AI use it for listings | Accuracy, fair housing language, final approval |
| Answering the first call or text | Picks up, answers questions, qualifies, books a showing | The conversation once the buyer is serious |
| Scheduling and follow-up | Books, reminds and follows up on old leads | Deciding which leads deserve your time |
| Home search | Matches listings to a plain-language request | What the photos leave out, and what the street is like |
| Pricing | Estimates values from comparable sales | Condition, the market this week, the pricing strategy |
| Negotiation | Can summarize offers | Reading the other side, and the counteroffer |
| Contracts and closing | Can summarize documents and deadlines | Licensed advice, inspection and appraisal problems |
The adoption numbers are real. NAR's 2026 technology report found 23 percent of Realtors use AI tools daily, 55 percent say AI has helped their business, and among those using it, 75 percent write listing descriptions with it and 52 percent draft emails and follow-ups. Real Estate News, covering the same survey, reported that only 21 percent of agents said they do not use AI at all, down from 32 percent a year earlier.
The portals are moving too. In October 2025 Zillow launched an app inside ChatGPT that answers home search questions with listings. Notice where it sends people once they find a home they like: to schedule a tour and connect with an agent. The search moved into AI. The transaction did not.
What AI cannot do for a buyer or seller
AI cannot hold a real estate license, owe a client a duty, or stand in a living room and tell a seller their price is 40,000 dollars too high. It cannot walk a house and notice the soft floor by the dishwasher. And it cannot carry a deal through a bad inspection, a low appraisal or a buyer who gets cold feet the night before closing.
- A license and accountability. The BLS notes every state requires brokers and agents to be licensed. Someone has to be responsible for the advice.
- Negotiation. Reading the listing agent, knowing when to push and when a seller is about to walk.
- Judgment on the property.Condition, the neighbor's barking dog, the road noise at 5pm.
- The bad days.Most of an agent's value shows up when something goes wrong between contract and closing.
There are legal limits on what AI can do for you, too. In 2024 the FCC ruled that AI-generated voices count as artificial voices under the TCPA, so an AI cold-calling homeowners needs the same prior consent as a robocall. Answering calls people place to you is a different matter, which is why inbound is where AI fits most easily.
An example of AI taking a task, not the job
Here is a concrete case, and it is ours. callflo.ai is an AI receptionist. It answers an agent's calls when the agent cannot: during a showing, at a closing, at dinner, on Sunday morning. It answers questions about listings from the agent's website, asks the qualifying questions, books the showing on the calendar and texts the agent a summary. It does not negotiate, give pricing advice, or pretend to be the agent. It hands the caller to the person who does that.
Sample call: a buyer calling during a showing
That is the realistic shape of AI in real estate for the next few years: software takes the call the agent would have missed, and the agent takes the conversation that decides the deal. The callflo.ai page for agents shows how it handles a listing line, and starts at 39 dollars a month on annual billing. For the wider toolkit, the best AI tools for real estate agents covers the rest, and the guide to answering calls during showings covers the hour you are with someone else's client.
Hear AI take the call you would have missed
Call callflo.ai's own AI agent and ask it anything about your business. When you are ready, pick a plan: no setup fee, no contract, and you do not pay unless you are satisfied.
Starter is $39 a month on annual billing. Live on your line about 5 minutes after you sign up.
What the job numbers say
The government's forecast is flat, not falling. The Bureau of Labor Statistics counts 530,600 real estate broker and sales agent jobs in 2025 and projects 2 percent growth from 2025 to 2035, slower than average, with about 40,400 openings a year as people leave and are replaced. Median pay was 52,830 dollars for sales agents and 73,220 dollars for brokers in May 2025.
NAR membership has been shrinking, and it is easy to read that as AI. NAR's own economists tie it to the housing market: when sales slow, part-time and low-volume agents leave, usually 18 to 24 months after the market turns. That is the same pattern NAR saw after the Great Recession, long before any chatbot. For what those first years look like, see how hard the job is for new agents.
What buyers and sellers actually want
Buyers use AI, but they do not trust it with the decision. Cotality's 2026 survey found 55 percent of homebuyers use generative AI at least monthly, yet trust in AI tools for finding a home fell from 30 percent in 2025 to 16 percent in 2026. 68 percent said they would verify most of what an AI told them, and 44 percent would pay extra for a human to check it. A NerdWallet survey run by The Harris Poll found about 48 percent of people planning to buy within a year intend to use AI along the way.
Put together: buyers will ask AI first and an agent second. Which means the first conversation with a human matters more, not less. In NAR's 2025 profile, 26 percent of buyers first contacted their agent by phone, more than any other channel, and 74 percent interviewed only one agent.