AI for real estate agents: where it helps, and what it can't hold for you

Every showing produces three things you'll need on Thursday and one place you'll forget them. This guide covers the drafting work AI does well for agents, the parts that should stay yours, the confidentiality question, and a simple way to keep what each client said where you can find it again.

Short answer: a general chat assistant is a good first-draft writer for listing descriptions, follow-ups and market summaries, and a fair summarizer of long documents. It is not a substitute for your judgment on price, contract terms or what an ad implies, and it knows nothing about what a buyer told you at a showing unless you tell it.

That second gap is the one that costs deals. After a Saturday of showings you're carrying what each couple said about the school catchment, what they winced at in the kitchen, and the question you promised the listing agent you'd ask. By Thursday, when they want to write an offer, it's in your head, three texts and a notebook in the car. Below: where AI is genuinely worth using, where it isn't, and a way to keep the client record that works whatever tool you use.

How agents are using AI in 2026

The best current numbers come from the National Association of REALTORS®. Its 2026 REALTORS® Technology Report (released 22 September 2026) finds that "Nearly half of agents now use AI daily (23%) or weekly (25%), and just 12% say they are not using it and have no plans to." Among agents who use it, the top jobs are:

• Writing listing descriptions: 75%
• Social media posting: 56%
• Creating emails and follow-up: 52%
• Market summaries: 30%
• Drafting marketing content with a personal tone: 30%
• Document review and summarizing: 27%

Two other lines from the same release are worth keeping in mind. Agents' main reason for adopting any technology is time: "Eighty-one percent said saving time is their primary goal." And the tools themselves are mostly general ones. The report's category is "AI-generated content (ChatGPT, Copilot, Gemini, etc.)", used by 41% of respondents, alongside the MLS (96%), e-signature (79%) and showing scheduling tools (68%) agents already rely on.

Where AI genuinely saves an agent time

These are the jobs where a general assistant earns its place, with the check that goes with each one.

• A first draft of the listing description. Give it your own notes from the walkthrough (rooms, updates, the light in the back bedroom at four o'clock) and ask for two versions in different lengths. You still edit it: every feature named has to be true of this house, and the tone has to sound like you.
• Follow-up emails after showings and open houses. Drafting a warm, specific note to eight open-house visitors is exactly the kind of repetitive writing it handles well, if you tell it what each person actually asked about.
• Market summaries from numbers you supply. Paste in the figures from your own MLS or CMA export and ask for a paragraph a seller will understand. Don't ask it to supply the numbers; a chat assistant is not your market data.
• Plain-English summaries of long documents. A 60-page HOA packet or a dense disclosure can be summarized into the five things a buyer will ask about. Read the original before you advise on any of them; the summary tells you where to look, not what to say.
• Getting the week's writing out of the way. Social captions, a newsletter paragraph, a price-reduction announcement. It is the category NAR's numbers show most agents already use.

What to keep in your own hands

The pattern across all of these: AI drafts, you answer for it. A few areas deserve more than a quick proofread.

• The facts in anything you publish. NAR's Code of Ethics (2026 edition, effective 1 January 2026), Article 12: "REALTORS® shall be honest and truthful in their real estate communications and shall present a true picture in their advertising, marketing, and other representations." A drafting tool that invents a "newly renovated" kitchen doesn't carry that duty. You do.
• Wording about neighborhoods and who a home suits. Descriptions of the area, the schools or the "ideal buyer" are where ad copy most easily says more than you meant. No tool makes that call for you; reread those lines with the fair housing training NAR requires of members every three years in mind, and ask your broker when unsure.
• Price and strategy. A summary can organize comparable sales. The recommendation to a seller, and the reasoning behind it, is the professional judgment the client is paying for.
• Contract terms. Article 9 of the same Code asks agents to make sure agreements are "in writing in clear and understandable language expressing the specific terms, conditions, obligations and commitments of the parties." Use your brokerage's forms and counsel, not a chat window, for the terms themselves.
• The relationship. Negotiation, the difficult phone call, the moment a first-time buyer loses their nerve. Nobody's clients want those handled by a draft.

Before you paste client details into a chatbot

Start with your obligations, then the tool's settings. NAR's Standard of Practice 1-9 says "The obligation of REALTORS® to preserve confidential information (as defined by state law) provided by their clients" continues after the relationship ends, and bars revealing it or using it to the client's disadvantage. What counts as confidential is set by state law, and many brokerages have their own policy on AI tools, so ask. Then, practically:

• Share what the job needs. A listing draft needs the house, not the seller's reason for moving or their bottom line.
• Check the training setting. OpenAI's help center (read 23 September 2026) says that when "Improve the model for everyone" is off, "your new conversations won't be used to train OpenAI models." The path: on the web, open your account menu, select Settings, select Data controls, turn Improve the model for everyone off and select Done. On iPhone or Android, open the sidebar, select your profile icon to open Settings, select Data controls and turn the same setting off.
• Know what a work plan changes. The same page: "By default, OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train its models." If your brokerage provides a workspace, use it for client work.
• Temporary Chat for one-offs. OpenAI says temporary chats "do not appear in your chat history," "do not create or update memories," "are not used to improve OpenAI models" and "may be retained for up to 30 days for safety purposes."
• Turning off training doesn't delete anything. OpenAI: "Turning off Improve the model for everyone does not delete or hide saved chats." Delete a chat separately if it shouldn't stay.
• Feedback is the exception. Thumbs-up/down feedback may still send that whole conversation to training. OpenAI: "If you choose to provide feedback, such as selecting thumbs up or thumbs down on a response, the entire conversation associated with that feedback may be used to train OpenAI models."

The part drafting tools don't solve: what each client told you

Here's the honest limit. A chat assistant can remember more than it used to. OpenAI's help center says ChatGPT's memory can draw on "past chats", "saved memories", "files in Library" and "content from connected apps, such as Gmail", depending on plan and region. But it also says "Memory does not retain every detail from every conversation. ChatGPT decides which available information is relevant to a response." And reaching your email that way means connecting a whole account, not the one deal thread you care about.

More to the point for an agent: the best information in a transaction is said out loud. "She really wants the walk to the elementary school." "He's worried about the roof but won't say it in front of her." "Their lease ends on the 30th, not the 15th." None of that is in a chat or an inbox. It exists in your memory for about a day, unless you put it somewhere.

A debrief habit that works with any tool

This works on paper, in your phone's notes app, or in anything else you'll actually use. The point is the habit, not the product.

• Sixty seconds in the car after every showing. Three lines: what they said (in their words), what they reacted to, what you owe them or anyone else. Say the address and their names so you can find it later.
• One place, not three. Debriefs scattered across texts, a notebook and your voicemail are as good as lost. Pick one home for them.
• One email thread per deal. Keep the lender, the title company and the other agent on a single running thread where you can. It's easier to reconstruct a timeline from one thread than from nine.
• Read the last debrief before the next contact. Two minutes before a call or a second showing, reread what they told you last time. Clients notice when you remember.

Or hand the file to a personal assistant

Equerry is a personal assistant on your iPhone that holds that record for you, from the things you send it. It works the way a good human assistant would: you tell it what happened, you copy it on what matters, and you ask it later.

• Debrief by voice after the showing. Dictate a voice note in the Equerry app on the way back to the car. It's transcribed, so what the Hendersons said about schools becomes something you can ask about on Thursday.
• CC it on the transaction thread. Your Equerry has its own email address. Add it to the deal thread and, for as long as it stays on the CC line, later replies from the lender or the other agent keep arriving. Or forward just the one email you need. It never asks for your inbox, and the original stays in your own mailbox.
• Send it the documents. Share an inspection report or disclosure PDF from another app with the share sheet, or save a photo of the handwritten notes from the listing appointment.
• Start the day briefed. If you allow calendar access, Equerry keeps a snapshot of your events from today through the next 14 days (titles, start and end times, locations and attendee display names, never event notes, meeting links or email addresses), and a Morning Brief, if you set one up, pulls together what you've saved that bears on the day's appointments, so you walk into the second showing knowing what they said at the first. Showing titles and locations often name clients and addresses, so include the calendar in the question you put to your brokerage.
• Ask in plain words. "What did the Hendersons say about schools?" "Where did we leave the repair credit on Maple Street?" When an answer draws on a note you saved, it carries a source chip: tap it to see the passage Equerry used and open the original.
• You can see what it kept. Settings → Your memory lists the facts it has kept, and you can delete any line.

Honest notes. Equerry doesn't connect to the MLS or to your CRM, doesn't join showings or record calls, and doesn't send email on your behalf; it knows what you send it, plus your calendar if you allow it, and nothing else. A CC is visible to everyone on the thread, so if the other side seeing an extra address is awkward, forward the email instead. And sending something is still sharing it: what you send is processed to make it searchable and answerable, and the calendar snapshot, if you allow it, is used to write your brief; both reach the outside providers named on our AI disclosure page. The confidentiality question above applies here too, so check your brokerage's policy first.

If the client record is the problem you came here to solve, the next reads are how to keep track of client conversations and how to keep a record of client calls.

Last updated 2026-09-25.

All product names, logos, and brands are property of their respective owners. Equerry is not affiliated with, endorsed by, or sponsored by any other product mentioned; comparisons are provided for informational purposes.