ChatGPT for Real Estate Agents: 12 Prompts That Work (and 3 Jobs to Never Give It)

Almost every “ChatGPT prompts for real estate” list has the same problem: the prompts are empty. “Write a listing description for a beautiful 3-bedroom home” produces exactly what you’d expect — a beautiful 3-bedroom home, described the way ChatGPT describes every beautiful 3-bedroom home, in copy that reads like the other four listings on the street that were also written by ChatGPT.

The prompts that work do one thing differently: they hand ChatGPT facts it could not have guessed. Square footage, the year the roof was replaced, the specific school, the actual commute time, the reason the sellers are moving. Everything below is built that way. Copy them, replace the bracketed parts with your real details, and the output stops sounding like a template.

Then there’s the part the prompt lists skip: three jobs where using ChatGPT will cost you money or get you in trouble. Those are at the bottom, and they matter more than the prompts.

Before anything: the Fair Housing problem

ChatGPT does not know US Fair Housing law, and it will happily write copy that violates it. Ask for a listing description and it may hand you “perfect for a young family,” “quiet neighborhood ideal for retirees,” or “walking distance to churches” — all of which describe who should live there rather than what the property is, and all of which are the kind of language the Fair Housing Act prohibits in advertising.

The Fair Housing Act bars advertising that indicates a preference or limitation based on race, colour, religion, sex, disability, familial status or national origin. “Family-friendly” and “perfect for singles” are the two that catch agents most often, and it does not matter that the AI wrote it — you published it.

So put this line at the end of every listing-related prompt you use:

Describe only the property and its features. Never reference or imply the age, family status, religion, race, nationality, disability or income of buyers, renters, or the people who live in the area. Do not use “family,” “family-friendly,” “bachelor,” “perfect for,” “safe,” “exclusive,” or “walking distance to churches.”

Every prompt below already includes a version of it. If you use nothing else from this page, use that.

Listing copy

1. The listing description that doesn’t sound generic

Write a 150-word listing description for a [property type] at [address/neighbourhood], listed at [price]. Facts: [beds/baths], [square footage], [lot size], [year built]. Recent work: [e.g. roof 2023, kitchen 2024, new HVAC]. Standout features: [3–5 specifics]. Nearby: [specific named places — the actual park, the actual grocery store, the actual transit line]. Lead with the single most unusual feature, not the bedroom count. Describe only the property; never reference buyers’ family status, age, religion, race, nationality, disability or income, and avoid “family-friendly,” “perfect for,” “safe” and “exclusive.” Plain, concrete language. No superlatives, no “nestled,” no “boasts.”

Banning “nestled” and “boasts” is not a joke — those two words plus “stunning” are the fastest way to make a description read as machine-written.

2. Three headline options for the MLS

Give me 3 MLS headline options, each under 80 characters, for [property]. One should lead with the standout feature: [feature]. One with the location: [neighbourhood]. One with the value angle: [price/condition]. No exclamation marks, no all-caps, no “must see.”

3. Rewrite for a different buyer, without changing the facts

Rewrite this listing description for a buyer who cares most about [e.g. an easy commute / a workshop space / low maintenance]. Same facts, same length, different emphasis. Change nothing factual and add nothing that isn’t in the original. [paste description]

4. Shrink it to fit the character limit

Cut this to under [n] characters without losing the roof year, square footage or the [key feature]. Keep whole sentences. [paste description]

5. A caption that isn’t “Just Listed! 🏡”

Write an Instagram caption for this listing: [3–4 real details]. Open with a specific observation about the property, not “Just listed.” Three to five short lines, one emoji maximum, then 5 hashtags mixing [city] and property type. Describe the property, not the buyer.

6. A 30-second video script from the photos

Write a 30-second walkthrough script for a listing video. Shot list: [describe your 6–8 photos in order]. One sentence per shot, spoken aloud in about 4 seconds each. Start with the most striking room, end with the address and a call to book a viewing. Conversational, no jargon.

7. Turn one listing into a week of posts

From these listing details, give me 5 different social posts that don’t repeat each other: one about a single feature, one about the neighbourhood with a specific named place, one about the price in context of [local market fact], one behind-the-scenes about preparing the home, one open-house announcement for [date/time]. Details: [paste].

Client communication

8. The follow-up that gets a reply

Write a short follow-up email to a buyer who viewed [property] on [date] and said [their actual reaction]. Reference their specific comment. Offer one concrete next step. Under 120 words, no pressure, no “just checking in,” no “I wanted to reach out.”

9. The price-reduction conversation

Draft talking points for telling a seller their home needs a price reduction. Facts: listed at [price] for [n] days, [n] showings, [n] offers, comparable homes at [comps] sold in [n] days. Lead with the data, not the recommendation. Give me three likely objections and a factual response to each.

10. Neighbourhood copy that isn’t a Wikipedia summary

Write 120 words about [neighbourhood] for a listing page. Use these specifics: [named coffee shop, park, transit stop, grocery store, commute time to downtown]. Describe amenities and distances factually. Do not characterise the people who live there, and do not use “safe,” “up-and-coming,” “family-friendly” or “desirable.”

Winning the listing

11. The listing presentation narrative

Write the marketing-plan section of a listing presentation for [property type] at [price point] in [area]. My plan: [list what you actually do — photography, single-property website, portal syndication, video, social, email, open house]. For each, state in one sentence what the seller gets from it. Factual and specific, no marketing adjectives. Under 250 words.

12. Your own bio

Write a 140-word third-person bio for a [title] in [market] who specialises in [specialties], with [experience level] and this background: [previous career, credentials, languages, local ties]. Use only these facts — invent no sales figures, awards or years of experience. Warm but professional.

If you’d rather skip the prompt entirely, our free real estate agent bio generator does exactly this, including a mode for agents with no closings yet. There are also free generators for listing descriptions, social captions and buyer emails — all with the Fair Housing rules built in, so you can’t forget them.

Three jobs to never give ChatGPT

Anything where a fabricated number is a liability. ChatGPT will produce comps, tax figures, HOA fees, square footage, school ratings and days-on-market that look completely plausible and are invented. It has no access to your MLS. If a number will appear in front of a client or on a listing, it comes from your MLS or your county records, not from a chat window. A confidently wrong square footage in an ad is a real problem, and “the AI said so” is not a defence.

Legal, disclosure and contract language. Purchase agreements, disclosure forms, addenda, lease clauses. These are state-specific, they change, and the failure mode is expensive. Use your brokerage’s forms and your attorney.

Anything that will be presented as a photograph of the property. We ran one empty room through ChatGPT’s image model and three purpose-built staging models: ChatGPT produced a beautiful room, but not the same room — windows re-proportioned, a radiator gone, the whole photo forced to a different aspect ratio. MLS rules prohibit altering walls, windows and dimensions, so the failure is a compliance one and it is invisible to you.

Distribution. This is the one that actually costs agents money, because it’s invisible. ChatGPT will write you a beautiful listing description in nine seconds — and then it sits in a chat window. It cannot put your listing on Zillow, build the property’s own website, cut the video, post to your channels, or get the page ranking for the address. Writing was never the bottleneck. Getting the listing in front of buyers and renters is the bottleneck, and no prompt solves it.

That’s the honest limit of ChatGPT in this job: it’s a fast, cheap first draft for anything made of words, and it does nothing at all about where those words end up. Reallyo takes one listing and produces the description, the video, the social posts, the single-property website and the portal syndication from a single upload — the writing plus the part that actually gets it seen. Use ChatGPT for the draft; don’t expect it to market anything.

A note on sounding like everyone else

By 2026 a large share of listing copy on the major portals is AI-written, and it shows — the same rhythms, the same adjectives, the same three-part sentences. That’s an opportunity, not a threat. The prompts above force specifics into the output precisely because specifics are what nobody else’s copy has any more. A description that names the actual roof year, the actual bus route and the actual coffee shop will stand out on a results page full of “stunning” and “nestled,” and it’ll be the one a buyer remembers.

If you want to know whether AI engines themselves are recommending you, that’s a different question with a real answer — our AI visibility tracker checks whether ChatGPT names you when someone asks it for an agent in your market.

FAQ

Is ChatGPT good for real estate agents? It’s good for first drafts of anything made of words — listing descriptions, captions, follow-up emails, video scripts, bios and listing-presentation copy. It’s unreliable for anything factual, because it has no access to your MLS and will invent comps, square footage, fees and school ratings that look plausible. Use it to write, then verify every number yourself.

What are the best ChatGPT prompts for real estate? The best prompts supply facts ChatGPT could not guess — exact square footage, the year the roof was replaced, named local landmarks, the actual commute time — and constrain the output with a length limit, a banned-words list and a Fair Housing instruction. Generic prompts like “write a listing description for a beautiful home” produce generic copy because there is nothing specific in them to work from.

Can ChatGPT write a listing description that complies with Fair Housing? Only if you tell it to. ChatGPT does not apply US Fair Housing law by default and will produce phrases like “perfect for a young family” or “quiet neighborhood ideal for retirees,” which indicate a preference based on familial status or age. Add an explicit instruction to describe only the property and never reference buyers’ family status, age, religion, race, nationality, disability or income — and read the output before publishing, because you are responsible for it, not the AI.

How do real estate agents use AI in 2026? Mostly for content drafting (descriptions, social captions, emails, video scripts), for summarising documents, and increasingly for lead response — an AI chatbot answering enquiries instantly rather than hours later. The parts AI does not solve are distribution and data accuracy: getting listings onto portals and in front of buyers, and anything requiring live MLS figures.

Will AI replace real estate agents? No, and the reason is visible in what the tools can’t do. AI can produce marketing copy in seconds, but the job is negotiation, pricing judgement, local knowledge, managing a transaction through inspection and financing, and being accountable when something goes wrong. What AI changes is the cost of the marketing layer — which means the agents who use it well spend their time on the parts that actually need a person.

Sources

Fair Housing advertising rules: HUD, Fair Housing Act and the National Association of REALTORS® Fair Housing guidance. Prompts were written and tested against ChatGPT in September 2026; model behaviour changes frequently, so treat every factual output as unverified until you check it against your MLS or county records.