The Prompt That Gives You Something to Say on an Expired Call
The script you have is the one every other caller is reading today. Ninety seconds of work on that one house gives you something none of them are holding.
105 CURATED Takes.
The script you have is the one every other caller is reading today. Ninety seconds of work on that one house gives you something none of them are holding.
Some clients distrust AI, and others run every document you send through a chatbot. Two questions at the first meeting tell you which kind you have, before either one costs you the deal.
You re-explain your market, your price points, and how you write at the start of every single chat, and you assumed that is simply how the thing works. It is a setting, it is two switches on one screen, and there is a third screen where you can read and edit everything it has stored about you.
Asking AI to write five tips for buyers gets you something clean, correct, and identical to what thousands of other agents just published. The fix in this piece is not a better instruction — it is reversing who is asking the questions.
Right now local posts happen in bursts and then stop for six weeks, because finding something worth saying depends on you remembering to look. Naming eight sources takes an hour, once.
You tried AI staging once, got back a room with the couch still sitting in the middle of it, and decided the tool was not ready. It was a wording problem rather than a capability one, and the fix is not a longer list of objects — it is one word about where.
AI listing video usually means handing a tool your footage and watching it quietly redraw the house until a window moves and the room is not the room anymore. The move here is the reverse: nothing in the original is allowed to change, and the thing being added has to obey the light that was already in the shot.
You assume Canva's bulk data tool will fill in the template you already built, and that it keeps failing because you set something up wrong. It ignores every text element that existed before you opened the app, which is why the field never links and why almost nothing written about this feature mentions it.
A buyer may walk in with a budget and a shortlist that an AI tool gave them. One question at intake tells you which of their numbers still need checking.
You assume connecting your CRM to an AI assistant is a question of picking the right middleware and paying somebody to wire it together. There is no connector to click at all, and whether any of it can work is decided by a single page that either exists inside your account or does not.
The danger with AI on a listing is not the obvious nonsense you would catch on the first read. It is one confident sentence about square footage or a school that nobody can trace back to anything, and the reason models produce those is that you never gave them anywhere to put uncertainty.
You already know an AI can summarize a contract, which is why it reads as a party trick rather than deal work. The seller net sheet is the part that matters at six in the evening when an offer lands, and whether you get one depends entirely on where you type the second question.
Handing software your email feels like the moment to lock everything down, so the strictest-looking switch is the one you reach for. That switch is the one that makes the assistant useless, and the setting you actually want protects you exactly as well.
You export a few columns from the MLS and call it comps. That is what the standard report gives you. The full export carries over 400 columns, and an AI reads all of them in the time it takes you to scan one.
A weekly note built from what happened around the listing does the pricing work for you. Here is what goes in it, and the part an assistant can build without you.
Sellers now turn up with a valuation they ran themselves and will defend. Arguing with it is what loses the room, and there is a move that points their own tool back at your plan.
Pulling an expired list is the same set of clicks every single time, and you have to be sitting there for all of them. On a live screen share, an MLS executive opened a browser extension on top of the MLS he was already signed into, said one sentence out loud, and got up.
Most agents treat AI staging as a money decision: pay a service per photo, or do it free in Gemini. Four sources in this packet treat it as two different decisions instead, which tool is allowed to touch a listing photo and whether that photo has to be labeled, and they do not put the line in the same place.
A 41-second phone clip went into an AI with one line typed, and a captioned Reel came back. The tool is free, and a few filming habits make the result noticeably better.
Every time you fix an AI draft somewhere else, the lesson gets thrown away. One short prompt hands it back, and the assistant starts asking why you made each change.
An agent who uses AI every day sorts every task with one question about the client. He also shows the two prompts he uses on work the client will see.
Meta has launched an AI tool for small businesses that can read your Instagram numbers, if you have access to it. One prompt turns them into what to keep, what to stop, and what to test next week.
An agent in the Atlanta area asked AI to read her own marketing back to her. The answer was not who she wanted to be, and a second question told her what to change.
An agent left a seller who was selling on his own, planning months of follow-up. His team sent one email instead, and the seller took his own listing down 20 minutes after it arrived.
An AI profile full of facts about you leaves the writing the same. Two choices in how the file is written decide whether the tool uses it and whether it changes anything.
Writing notes in a client meeting pulls your eyes off the client. Google's meeting app can now take them for you in person, and what it sends back afterward is the useful part.
Most agents describe how they write and get AI drafts that are only mostly right. The part that closes the gap is a list almost nobody makes, and your old posts already hold part of it.
By one AI company chief's account, the blank contract forms agents use are licensed to them, not owned. He names the setting that decides whether an AI tool learns from what you upload.
Google's own guidance tells websites to fact-check AI writing by hand before it goes live. The instruction reaches past the article itself, into parts of the page that are easy to forget.
A team leader who covers Houston news has AI read the city's planning filings and send her a summary. The setup works for any city, and it starts with finding where yours posts them.
One agent asked a consumer AI assistant for three homes in her own town and told it to compare them. What it said about her listing is what a buyer may hear first.
A paper form on your desk can be scanned, filed, and sent out for a signature without leaving Google Drive. A broker who trains agents shows each step.
You keep the deadlines in your head, or you type them into your calendar off the contract one deal at a time, and the option period is the one that eventually bites somebody. He connected an AI assistant to his email and calendar, spoke a single instruction naming one address, and it went and found the contracts, read them, and set the dates itself.
A Florida agent types a closing goal into a free chatbot and gets back three groups of sellers, in order. Two short instructions at the end of the prompt are the part worth copying.
An Atlanta agent stages his listing photos in ChatGPT with three separate requests and two extra uploads. He also has a firm rule for when to quit a chat and start over.
If your website runs on Real Geeks, a paid AI assistant can now write and change its pages for you. It starts with one web address, and the choices you make at setup decide what it can touch.
Your whole market is probably not stalled. A year of closed sales and one question to an AI tool can show you which corner is still moving, so you know where to spend your time.
You send market reports and share the link everywhere, and you assume the platform captures anyone who opens one. It captures nobody unless they click subscribe, and the control that changes it is three clicks deep under a line of small print in your own profile.
A creator built an AI panel that argues over every content idea before he films it. One seat is built to love the idea, one is built to kill it, and one makes the call.
Most agents give AI a property profile and a few photos and hope for the best. One agent builds a full file from seven places first, including screenshots that show the AI how the rooms connect.
A machine transcript can write a pool where the buyer said no pool. The CEO of Ditto Transcripts names the small words to check against the audio before a recap goes out.
One email each weekday morning can sum up your inbox, your day, and your deals. A title insurance rep who teaches agents to use AI lists the parts he built into it, and two of them tell you what to do today.
Claude can write a batch of social media posts and put each one on a posting calendar, dates and all. The setup takes a few clicks, and the scheduling tool has a free version.
One Central Florida agent has an AI tool read 41 local groups day and night. The setup fits in a few lines, and one line tells the tool never to speak for you.
One coach says a calling habit some agents recommend can keep your calls from getting through. A brand new number can be flagged on its very first day, too.
A Tucson agent built her whole pitch on one strength. Her favorite clients' reviews, run through an AI tool, praised something else, and the questions she asked are easy to copy.
One listing agent pastes the week's showing comments into AI and gets back a short email for the seller. One line in the prompt pushes back on AI's habit of softening bad news.
Most follow-up sounds the same every time. One coach's prompt gives each message its own reason to reach out, and one rule decides when you may ask for a meeting.
One agent had a new Meta tool study 1,133 old posts before asking it for a single idea. These copy-paste prompts show where your account is weak, who follows you, and where your time is going.
A property data app now writes a pricing report with AI. Its own pages show the inputs to check first, and the setting that shows your condition calls to the seller.
One request to an AI assistant turns a spreadsheet into a Word file that prints right on the envelope. This agent printed 38 in about 20 minutes.
A group leader now records every call and hands the transcripts to AI with a short request. The wording ties the ideas to what was really said, and there is one thing to check before you record.
You know you should text your past clients individually and you do not, because writing a personal message to two hundred people is a week you do not have. This team saved a prompt where a message template normally goes, and each text takes about twenty seconds to produce.
You think the question about AI and MLS data is whether you are allowed to use it. The technology executive on this podcast says the real question is which fields, and that the most common way agents get it wrong is not a chatbot conversation at all.
You answer let's start high and come down later with market talk and a feeling, and the seller has a feeling too, so the appointment ends in a draw. A coach had an AI assistant read his own MLS and plot days on market against percent of original list price, and in that data the first week held full asking price.
You have watched someone automate MLS work with an AI agent and assumed the only question is whether it works. An Austin agent found out that the speed is what gets noticed, and that his MLS responded by rewriting the end user license agreement.
Your AI-written follow-up sounds like a salesperson because you told it what to write and never told it who it was. A California agent adds one closing sentence to every prompt she writes, plus a line naming three phrases the AI is not allowed to use.
You have been sold AI voice as the thing that finally works your old database while you sleep, and you either bought it or you decided the whole category is nonsense. The company that spent about eighteen months and roughly fifty thousand leads testing it abandoned the replacement model, then watched a client succeed with the same technology by changing one thing.
Advice on making a listing video look good has always been about lighting, angles, and editing. These three sources are talking about the same thirty seconds of footage and each wants something different put in - or taken out.
You already know AI can summarize comparable sales, so you have tried it once and gone back to doing the analysis by hand. This coach built a seven-page instruction set, pointed a browser agent at his own MLS, and got back a finished report, a page-by-page talking-points document, and a custom graph for the listing appointment.
You have been told to set up an AI assistant for your inbox and calendar, and the setup guides all stop at connecting it. The two sentences that decide whether it is useful or dangerous go in the project instructions, and they are short enough to type from memory.
Your profile bio was written for a person who is already looking at your page, and most agents have not touched theirs in years. The rewrite she demonstrates puts two or three short, specific paragraphs on top and then a set of questions phrased the way a stranger would type them into a chatbot.
You judge an AI assistant's morning briefing by whether it looks complete, and a complete-looking briefing feels like it worked. A format with sections in it creates pressure to fill every section, and the day nothing happened is the day it writes fiction.
The pitch going around is that you can cancel most of your software and rebuild it yourself for a fraction of the cost. The people making that pitch draw one line themselves, and it runs straight through the system your entire business sits on.
You assume nobody can tell which of your posts you wrote and which one a model wrote for you. There is one sentence pattern that gives it away in the opening paragraph, and consumers already told a survey that the whole industry sounds the same.
You think of your software as a stack you chose deliberately, one tool at a time, over several years of building the business. The speaker on this clip says most agents are quietly paying for ten to twelve subscriptions, that most of them can go, and that finding out takes one sitting with a credit card statement.
The fifteen minutes before school pickup are the ones you write off, because nothing worth starting can be started and finished in fifteen minutes. An agent typed that exact constraint into her CRM's AI assistant and came out of it with three named contacts and three drafted emails, sent before she left the office.
An AI product builder was told no one at work put company or client data into free AI tools. Then she ran a survey with no names on it, and the firm found it needed tighter controls.
Green-screening a news article is already what half your feed does, and it works because you are commenting on something real rather than announcing your own listing. The tool described here removes the only hard part, which is writing the version of that story that is about your town.
Agents worry about AI detectors and watermarks, but the thing that actually outs a post is a reader who recognizes the rhythm. A famous investor's op-ed got flagged in public by ordinary readers, and it came down to one sentence construction and one punctuation mark.
The debate about AI assistants is usually all or nothing: either you hand over the keys or you stay out. This broker has gone further than almost anyone in giving an assistant real access, and she draws one hard line that has nothing to do with how much she trusts the software.
Every AI-written listing article comes back with the same three features: great location, nice kitchen, big yard. Those are the fields in the MLS, not the reasons anyone buys, and this prompt bans them outright with one line most people never think to include.
You search your own name, find yourself on page one, and file the problem as handled. The pitch behind this tool is that most agents do fine on their own name and vanish on the searches that actually produce a client, like "best real estate agent near me".
You assume that once you paste a drafted description into the MLS it is simply your text, indistinguishable from anything else you wrote. Since August 2026, text from one of the major assistants carries a pattern that says otherwise, and pasting does not remove it.
Somebody sent you a skill file, it landed in Downloads with an extension you have never seen, and that is where it has stayed. The menu path is four clicks, and the more useful half is that you can build your own by talking.
The comfortable assumption is that AI can draft your emails but should stay nowhere near a live file with dates on it. One team leader says she replaced her stateside staff and full-time virtual assistants with a single AI coordinator. It tracks files and works with title and the lender.
You have been telling priced-out buyers that a low rate is something other people locked in years ago. Some of those loans are still attached to homes sitting on the market right now, and nothing on the listing page tells you which ones.
An automated reply sitting under a fake review does the exact thing you least want, which is to point at it and make it look like a live dispute is happening. Turning on spam detection pulls those reviews out of your widget, your dashboard, and your bot's reach, and it does none of that on Google itself.
The processes that eat your week are the ones living in your head that have never been written down anywhere, which is exactly why you have never automated them. You can record your screen while doing the job at normal speed and talking through it, and the recording becomes the instructions.
You solve the same problem with a fresh prompt every month, spend twenty minutes refining it, then lose it somewhere in your chat history. One sentence at the end of a conversation you were already having turns that work into something you can run again, with no file and no code.
ChatGPT keeps handing back listing copy you legally cannot use. The fix is not a better prompt.
Anthropic shipped workflows that plug an AI into QuickBooks, HubSpot, Docusign, and Drive. The reassurance everyone quoted has a scope written into the sentence.
One real estate team's best-performing posts are not videos and carry no branding at all. What the team stopped asking ChatGPT for is the reason the posts get forwarded.
An agent published his own ad account: $1,300 spent, 170 clicks, 13 seller leads. It is one unoptimized account in one market, and it is the first real-money number anybody has shown.
Agents quietly build workarounds for the chatbot that stops mid-answer on mold, radon, or a buyer's medical accommodation. The vendor has now published how much of that it removed.
AI writes like a stranger because every new chat starts with nothing. The permanent fix takes one afternoon, and the files that make it work are ones you already have.
Most agents ask AI for a list and take whatever comes back. One sentence tells it how much work to do, and it changes what shows up.
Agents buy tools built on MLS data assuming every version of it holds the same thing. It does not, and the tier that is easiest to get holds no record of what anything sold for.
Almost half of agents who have tried AI say it changed nothing for them. The reason is in what they asked it for, and one listing shows the difference.
Automations do not break loudly. They stop, and new leads keep arriving into nothing, which is how four days of them were lost before anyone noticed.
Anyone can now generate a market script that sounds like every other agent's. The fix is not a better script, and it costs you nothing you do not already have.
An Austin agent has closed deals this year off videos he never sat down to film. What keeps that from costing him anything is one line of disclosure, and where he puts it.
Feeding AI samples of your writing is only half the job. One step in between shows you what it got wrong, before a single draft goes out under your name.
One paste returns the list, the fees, and the applications already filled in. What decides whether any of it turns into a closing happens after you send them.
Switching tools normally means starting over, so most people keep two and explain themselves twice. There is a built-in screen that carries it across in about a minute.
Part of your inbox needs a reply and no thought. Name that part out loud, and an assistant that has studied how you write can have the drafts waiting for you.
The receipts pile up all year and get keyed in at the worst possible week. He points the tool at the folder instead, and one half of his instruction decides whether the result is usable.
You cannot tell whether a tool saved you anything, because the time you spend fixing its work is never written down. One month of counting settles it.
Most tool trials never end. They fade, because nobody wrote down what winning would look like. One sentence and one date decide it for you instead.
Your written instructions move to the new format on their own. One part does not, and the links you already handed out could break for the people who use them.
The notes you already typed hold the whole update. A prompt you can paste turns them into one page, and one line in it stops a clean-looking draft from hiding a guess.
You paste in a sample you like and the result still comes back flat. There is a second half to that move, and it is one sentence long.
Your buyer sends pictures of houses, not addresses, and somebody has to find every one of them in the MLS. One agent hands the whole batch over at once.
Most agents hand AI the topic and ask for the article. This brokerage puts one step in front of the draft, and it asks nothing of you except answers.
A daily automated report is easy to set up and easy to stop reading, and it keeps billing you the whole time. Two changes keep the ones worth keeping alive.
© 2026 CURATED For Agents. A CURATED Takes Brand.
3343 Peachtree Road Northeast, Atlanta, GA
Privacy · Terms · support@curatedforagents.com
This one is part of CURATED Pro.
CURATED Pro opens all of this:
We read 7,380 real estate sources in the last 24 hours
and saved you 227 hours.
7 days free, then $21/month. Cancel anytime.
After that, CURATED Pro is 7 days free, then $21/month.
Every script, tactic, and story worth knowing, in one place.
Every tool, the Library, Ask CURATED, and every prompt.
7 days free, then $21/month while launch pricing lasts, less than 70 cents a day.
Cancel anytime.
Get CURATED free in your inbox, every weekday.
Free.
Unsubscribe anytime.
Get CURATED free, every weekday.
Free.
Unsubscribe anytime.
Get CURATED free in your inbox, every weekday.
Free.
Unsubscribe anytime.