The short answer
Real estate agents get the most from AI on listings, marketing and follow-up: listing descriptions and lead replies with ChatGPT or Claude, flyers and social graphics in Canva, local research with sources in Perplexity, and translation with DeepL. Keep pricing advice, negotiation and disclosures human. In the US, the Fair Housing Act bans adverts that indicate a preference based on protected traits, and HUD's withdrawal of its 2024 digital advertising guidance, published in April 2026, did not change that law.
| Start with | ChatGPT or Claude for listings, Canva for marketing |
|---|---|
| Always run | A fair housing check on every AI-written listing |
| Keep human | Pricing advice, negotiation, disclosures, ad targeting |
| US law | Fair Housing Act, 42 U.S.C. 3604(c), covers all housing adverts |
| HUD guidance | 2024 digital advertising guidance withdrawn effective 17 September 2025 |
How real estate agents are using AI in 2026
Real estate is a writing and follow-up business as much as a selling one. Every listing needs a description, every property needs social posts and a flyer, every lead needs a timely reply, and every client wants to know what the market is doing. AI handles the first draft of all of that well. What it does not handle is the law around housing advertising, the accuracy of what you say about a property, and the judgement that wins a negotiation.
A small kit covers most of the job. ChatGPT or Claude write listings, emails and market updates. Canva turns them into flyers, social graphics and open-house signs with your brand kit. Perplexity researches local facts with sources, and DeepL translates listings for buyers who read another language. A meeting assistant such as Fireflies.ai captures client calls, with permission.
The 7 real estate tasks worth handing to AI
1. Write listing descriptions from your notes (ChatGPT or Claude)
Give the model the facts: rooms, measurements, features, recent work, outdoor space and location details you have checked. Ask for a description that sticks to those facts and describes the property, not the people who should live there. Then run the fair housing check prompt below on every draft before it is published.
2. Turn a listing into social posts and a flyer (ChatGPT and Canva)
Ask ChatGPT for posts in three lengths, then build the flyer and graphics in Canva, which applies your brand fonts and colours. Use your real photos. Our small business stack on the stacks page shows the same ChatGPT and Canva routine for a week of posts.
3. Reply to new leads quickly (ChatGPT)
Speed wins leads. Keep a project with your listings, viewing process and common answers, and ask for a reply to each enquiry that answers the question and offers two viewing times. You check the details and send. Treat every lead the same way, whatever their name or how they write.
4. Write a monthly market update (Claude)
Paste your own figures from your MLS or market data provider, such as sales, days on market and price changes, and ask Claude for a plain-English update for past clients. It writes clearly, and telling it to use only the figures you gave it makes invented numbers less likely, but check every figure against your data before sending.
5. Research local facts, with sources (Perplexity)
Planned transport links, zoning changes, new developments and local rules all come up in conversations with clients. Perplexity gives a sourced answer you can check. Stick to facts about the area and the property; avoid describing neighbourhoods by who lives there.
6. Translate listings and client messages (DeepL)
DeepL produces natural translations, and its glossaries keep terms consistent. Use a paid plan for client messages, because DeepL says texts on the free version may be used to train its models while paid plans delete them after translation.
7. Summarise client calls and keep follow-ups on track (Fireflies.ai)
After a buyer consultation, a summary of their must-haves, budget range and timeline makes follow-ups sharper. Fireflies.ai says it never uses meeting content to train AI models. Ask permission before recording, every time.
What real estate agents should keep human
- Pricing advice. A valuation or list price recommendation draws on local knowledge and comparable sales you have checked. AI can lay out the numbers; you give the advice.
- Negotiation. Reading the other side and protecting your client's position is the job.
- Disclosures and contract terms. What must be disclosed and what a contract means are legal questions. Do not let AI summaries stand in for the documents or for a lawyer.
- Ad targeting choices. Who sees your adverts is a fair housing decision, not just a marketing one.
- Your relationships. Clients remember who called them back and who understood what they needed.
A starter AI stack for real estate agents
| Job | Tool | Plan to start on | Why |
|---|---|---|---|
| Listings, emails, market updates | ChatGPT or Claude | Free, then a paid plan | Fast, clean drafts from your facts |
| Flyers and social graphics | Canva | Free, then Pro | Brand kit and editable AI designs |
| Local research | Perplexity | Free, with AI data retention turned off | Sources you can check |
| Translation | DeepL | A paid plan for client messages | Natural wording; paid texts not used for training |
| Client calls | Fireflies.ai | Free to start | Says it never trains on meeting content |
For the marketing side in more depth, see the best AI for marketing and the best AI for social media, and for wording, our ChatGPT prompts for sales and email prompts.
A 30-day AI plan for a real estate agent
- Days 1 to 7: set up and check. Choose one writing assistant and switch off training on a personal account (details below). Create a project with your bio, your process, your common answers and two listings you are proud of. Read your brokerage's AI and advertising policy.
- Days 8 to 14: listings with a check. Write your next listings with AI, then run every draft through the fair housing check and your own read. Note what you changed.
- Days 15 to 21: marketing and leads. Build Canva templates for flyers and posts. Use AI drafts for lead replies and track your response time.
- Days 22 to 30: the client update. Send your first AI-assisted market update to past clients using your own figures. Review replies and enquiries, and keep the parts that worked.
The risks for real estate agents: fair housing and accuracy
This is a summary of official US sources, not legal advice; other countries have their own equality and advertising rules.
The Fair Housing Act applies to every advert, whoever wrote it. Section 3604(c) of the Fair Housing Act makes it unlawful to make, print or publish, or cause to be made, printed or published, any notice, statement or advertisement about the sale or rental of a dwelling that indicates a preference, limitation or discrimination based on race, colour, religion, sex, handicap, familial status or national origin. An AI-written listing is your listing. Models trained on old adverts can reproduce phrases that describe who the home suits rather than what it offers, so check every draft.
HUD withdrew its digital advertising guidance, not the law. In a Federal Register notice published on 6 April 2026, the US Department of Housing and Urban Development listed its April 2024 guidance on applying the Fair Housing Act to advertising through digital platforms among documents withdrawn effective 17 September 2025. The same notice says actions that do not comply with the text of the Fair Housing Act remain subject to HUD enforcement, and that complainants can still file a civil action in federal or state court within two years. Fewer guidance documents means you rely more on your own checks, not less.
Targeting is part of the advert. Choosing audiences for paid social adverts by characteristics linked to protected classes can create the same problem as discriminatory wording. Keep targeting to the property's location and price, and follow your brokerage's rules.
Invented or edited details mislead buyers. AI can add features a property does not have, or describe a commute it cannot deliver. AI image tools can make a room look different from reality. Check every fact against your notes, and follow your MLS and brokerage rules on edited or virtually staged photos.
Client data. Buyer budgets, mortgage details and personal circumstances are confidential. ChatGPT Free, Plus and Pro train on chats unless you switch off "Improve the model for everyone" under Settings, Data controls, and Perplexity keeps AI data retention on by default on Free, Pro and Max. Keep sensitive client details out of personal AI accounts.
Prompts to copy
Listing description from your facts
Write a listing description for this property using only these facts: [rooms, measurements, features, recent work, outdoor space, checked location facts]. 150 to 200 words, warm but factual, leading with the strongest feature. Describe the property and its features, not the type of person or household who should live there. Do not add any feature I have not listed.
You get: A factual listing draft
Fair housing check
Review this listing and list every word or phrase that describes or implies a preferred buyer or tenant, or refers to race, colour, religion, sex, disability, familial status or national origin, directly or indirectly (for example words about the ideal household, the neighbours or nearby places of worship). For each, explain the concern and suggest wording that describes the property instead. Listing: [paste]
You get: A list of phrases to fix before publishing
Social posts and flyer text
From this listing, write: one Instagram caption under 100 words, one Facebook post under 150 words, one short post under 40 words, and flyer text with a headline, five bullet features and a call to action for the open house on [date and time]. Use only the facts in the listing.
You get: Marketing copy ready for Canva
Reply to a new lead
Using the listings and answers in this project, reply to this enquiry. Answer their question directly, offer two viewing times ([times]), and ask one question about what they need. Under 90 words, friendly and professional. Enquiry: [paste, without personal details you do not need]
You get: A quick, helpful lead reply
Monthly market update
Write a monthly market update for past clients in [area] using only these figures: [sales, median price, days on market, inventory, changes versus last month and last year]. Plain English, under 200 words, one line on what it means for sellers and one for buyers, no predictions. Do not add any number I have not given you.
You get: A client email built on your real data
Local facts with sources
What planned transport, development or zoning changes affecting [area] have been announced in the last 12 months? List each with its status and a link to the official source (council, transport authority or planning portal). Do not describe the area's residents.
You get: Sourced local facts to mention to clients
Buyer consultation summary
From this transcript of a buyer consultation, summarise: must-haves, nice-to-haves, deal-breakers, budget range, timeline, financing status and agreed next steps. Under 150 words. Mark anything the client seemed unsure about.
You get: A clear brief for your property search and follow-up