ChatGPT Prompts for Real Estate Agents (Listings, Follow-Ups, and Content That Converts)

Real estate runs on writing nobody has time for: listing descriptions at 9 p.m. the night before photos go live, follow-ups to forty open-house sign-ins, the market-update email you've been meaning to send since March. ChatGPT drafts all of it in minutes — and it will cheerfully draft you straight into a fair-housing violation or an invented square footage if you use it naked.

This is the working prompt set for agents: the setup prompt, the listing machine, the follow-up sequences, the content batches, and — first, because it matters most — the three compliance rules that make AI drafts safe to send.

The three compliance rules (before any prompt)

1. Fair housing screens every draft. The Fair Housing Act's protected classes (race, color, religion, sex, disability, familial status, national origin — plus state/local additions) apply to your marketing language, and AI loves the exact phrases that trip it: "perfect for young families," "walking distance to churches," "safe neighborhood," "exclusive community." Describe the property, never the ideal buyer. Add this line to every marketing prompt: "Comply with the Fair Housing Act: describe property features only; no language that suggests a preferred type of buyer, and flag anything borderline for my review."

2. Facts come from you, not the model. ChatGPT will invent square footage, school ratings, HOA fees, and "recently renovated" with total confidence. Give it the spec sheet; forbid additions: "Use only the details I provide. If a detail is missing, leave a [BRACKET] — never fill in facts." You verify brackets against the MLS before anything ships.

3. Client data stays out. No names tied to financial situations, no pre-approval amounts, no divorce-sale backstories. Describe situations generically ("a seller relocating on a 6-week timeline") and personalize after the draft leaves the chat. Your brokerage's AI policy — and your state's advertising rules, including team/brokerage disclosure requirements — override anything on this page.

The setup prompt (paste once per session)

You're the marketing assistant for a real estate agent. My market: [city/area, price band, property mix]. My clients: [first-time buyers / move-up / investors / luxury]. My voice: [e.g., warm and plainspoken, no hype words, light humor, no exclamation points]. Rules: comply with the Fair Housing Act (property features only, never buyer types); use only facts I provide, [BRACKET] anything missing; keep everything skimmable on a phone. Ask up to 2 clarifying questions if a request is ambiguous; otherwise draft.

1. The listing description machine

Write a listing description for: [paste the raw spec sheet — beds/baths, sq ft, lot, year, upgrades with years, neighborhood name, notable features]. Give me 3 versions: (1) MLS, 800 characters, leads with the two strongest features; (2) 150-word narrative for the flyer/website with a headline; (3) 40-word teaser for social. No clichés ("must see," "won't last," "oasis"), no buyer-type language, active voice. Then list the 3 features you led with and why — so I can overrule you.

The "explain your lead" clause is the difference between an assistant and a slot machine: if it led with the kitchen when the lot is the story, you fix strategy, not sentences.

Price-reduction reframe: "Rewrite this listing copy for a price adjustment: new price [$X]. No apology language, no 'motivated seller.' Reframe around value at the new number and one feature we haven't emphasized yet."

2. Follow-up sequences that actually go out

The fortune's in the follow-up, and the follow-up is exactly what dies on busy weeks. Batch it:

Open house, same night:

Write 3 short follow-up emails for open house sign-ins at [address]: (1) serious buyers who asked questions — reference [the question topics], offer a private showing; (2) casual/neighbor visitors — no pressure, offer my monthly market email; (3) agents/vendors — brief and collegial. Under 90 words each, one clear next step, my voice. Subject lines included.

The long game: For leads that go quiet, adapt the cadence in our follow-up email guide — the "value per touch" rule matters double in real estate, where the sales cycle is measured in months and the unsubscribe is forever.

Past-client touches (where referrals live):

Draft 4 quarterly check-in emails for past clients for the next year. Each gives one genuinely useful thing — [home-maintenance seasonal reminder / neighborhood sales snapshot / property-tax or insurance deadline / an "your equity probably changed" note] — in under 120 words, warm but not salesy, ending with a soft referral line that doesn't beg.

Cold outreach to FSBOs/expireds: structure and personalization rules are the same as any cold email that gets replies — lead with their situation, one specific observation, one low-commitment ask. Have ChatGPT draft 3 variants and A/B them.

3. The content batch (a month in 30 minutes)

Create my content calendar for [month]: 12 posts across Instagram and Facebook for an agent in [market]. Mix: 4 educational (buyer/seller process myths), 3 market data posts (I'll paste stats — build the hook and caption around them), 2 neighborhood spotlights [names], 2 behind-the-scenes prompts I can film in 60 seconds, 1 client-win template (no client details). For each: hook line, caption under 150 words, CTA, and a filming/photo note. My voice, fair-housing safe, no "DM me before it's gone" urgency theater.

Feed the market-data posts real numbers from your MLS each month; the hooks and structure are reusable forever. For platform-specific polish, run finals through the frameworks in Instagram caption ideas and hooks and — for the professional audience that refers you business — LinkedIn posts that don't sound like AI.

The market update email (the one you keep not sending):

Turn these stats into my monthly market email: [paste: median price, DOM, inventory, list-to-sale ratio, YoY changes for your area]. Structure: one-sentence "what this means" up top for [buyers/sellers/both], 3 bullet takeaways in plain English (no jargon — translate "months of inventory"), one paragraph of local color I'll provide: [note], CTA offering a no-pressure equity or strategy chat. 250 words max. Do not editorialize beyond the data I gave you.

4. Client-facing explainers and negotiation prep

• The translator: "Explain [escrow / contingencies / earnest money / appraisal gap] to a first-time buyer in 4 sentences, a friendly analogy, and one 'what this means for you' line. No legalese."
• The objection rehearsal: "Play a [seller who thinks their home is worth $50k more than comps / buyer scared by rates]. I'll respond; push back realistically, then critique my answer — what was strong, what sounded defensive, what I should ask instead. Stay in character until I say stop."
• CMA narrative: "Here are the comps: [paste addresses, prices, DOM, adjustments]. Write the one-page narrative explaining my recommended range of [$X–$Y] — why each comp matters, adjustments in plain English, and the risk of overpricing stated kindly but plainly. Use only these numbers."
• Offer-summary sheet: "Turn these offer terms into a side-by-side summary a stressed seller can read in 2 minutes: [paste terms]. Flag meaningful differences beyond price (financing strength, contingencies, timelines) neutrally — I make recommendations, you organize information."

That last clause is worth internalizing as the whole philosophy: ChatGPT organizes; you advise. The license on the wall is yours.

Make it a one-click tool

If you're pasting the setup prompt daily, graduate to a Project or a personal custom GPT holding your market, voice, compliance rules, and best past drafts — then "listing description, spec sheet attached" is the entire prompt. Our guide to the best custom GPTs for productivity covers the pattern; agents are its heaviest users for a reason: same context, forty uses a week.

FAQ

Can I use ChatGPT for listing descriptions on the MLS? Generally yes — you're responsible for the output exactly as if you'd written it. That means fair-housing compliant, factually verified against the listing, and within your MLS's advertising rules. The prompts above build the first two in; your MLS handbook covers the third.

Will AI-written content hurt my brand? Generic AI content will — the "nestled in the heart of" voice is instantly recognizable and forgettable. The fix is the setup prompt's voice rules plus your local knowledge: AI structures the draft, your details make it yours. If a paragraph could appear in any agent's feed in any city, it's not done.

Is ChatGPT safe for client information? Treat the chat like a semi-public space: no names attached to finances, no confidential deal terms, nothing you wouldn't put in a group text. Generic scenarios in, personalization after. Check whether your brokerage requires business-tier AI accounts (which exclude training on your data) for anything client-adjacent.

What about Zillow/Redfin descriptions, are those different from MLS? Same discipline, different lengths — portals truncate hard, so ask for a front-loaded 400-character version where the first sentence survives the cut. The 3-version listing prompt above covers it; add "portal version: 400 characters, key facts in sentence one."

Do these prompts work in Claude or Gemini instead? Yes — the system (setup context, fact brackets, fair-housing screen, explain-your-choices) transfers to any capable model. The compliance rules are about your process, not the vendor; whichever tool your brokerage approves, the prompts port over nearly verbatim.