Train Your Team: AI Prompts for Agents (The Four-Part Brief)

Why I Taught Agents to Brief AI (and Why It Helps You Recruit)

Your agents are already using AI. The question is whether it's helping them.

Some get a listing description they can use in a minute. Others get 400 words that sound like every other AI listing on the MLS, decide AI doesn't work for real estate, and never open it again. Same tool. The difference is the brief they gave it.

When agents asked me about AI, I didn't start with theory. I started where they were. The way I describe it: "I would just get better and learn more and solve each agent's problem of what they felt like they needed."

What they felt like they needed was not a mystery. "Most of them just wanted to build their brand and learn AI to get more leads. So I built the things that would help them sell more houses." The prompt method below is the one I gave the agents in my coaching cohorts.

Here's why a team leader should care. Tools don't hold agents. People do. As I put it on a recent call, "Human beings leave other human beings. To be led. By other human beings." Teaching an agent to get real work out of AI is part of leading them.

It's also a recruiting tool. "This is the stuff I would teach and build so I could recruit more agents." Experienced agents get pitched software all the time. A team that can show them how its agents get usable work out of AI, in a live meeting, is offering something a software demo can't.

I'm Blake Suddath, a real estate recruiter and recruiting coach and the co-founder of Inner Cirql Recruiting Systems with Tyler Lewis. This lesson is part of the Team Training Library. It pairs with what to automate and what stays human and AI CRM setup that agents actually use. For the bigger picture of how this fits a growing team, start with the recruiting page.

How to Roll This Out to Your Team

You don't need an AI course. You need three places this already fits.

Teach it live at your weekly company meeting. Pick one task every agent has this week, like a listing description or a follow-up text, and build the brief on screen together. Then compare the drafts. In my own model that meeting was on Tuesdays, and the most success I had was Tuesday at 11am. I'd invite agents to "just come sit in the back, I'll save you a seat in the back, be a fly on the wall." A guest who watches your team turn a vague prompt into a usable draft is seeing exactly what they'd get.

Put it in onboarding. I'd onboard new hires in the morning and then they'd go straight into the company meeting. Add one job to that morning: each new agent writes their first brief for a real task, so their first week of copy doesn't sound like everyone else's.

Check it in your weekly one-on-ones. We book those for 45 minutes, and they usually run a little longer. Ask for one prompt the agent wrote this week and what they changed after the first draft. You're coaching a habit, not grading writing.

On a team running the Territory Takeover System, agents also join the weekly Always-On coaching call on Wednesdays, so the lesson gets reinforced outside your own meeting. The Inner Cirql explains that agent layer in what the Always-On Agent System is.

The Agent Lesson: AI Prompts That Produce Usable Copy

Everything from here down is written to your agents. Print it, share the link, or teach from it.

You don't have an AI problem. You have a brief problem.

Most agents type one sentence, like "write a listing description for my new home in Edina." They get back a block of text that sounds like every other listing on the MLS. They decide AI doesn't work for their business. The tool is fine. The brief was missing.

AI works with what you give it. Give it one sentence and you get one sentence of effort. Give it a brief and you get a draft you can actually use.

The RICE Brief: Role, Input, Constraint, Example

Every prompt that works has four parts. Role, Input, Constraint, Example. Call it RICE.

Role. Tell the AI who to be, in one sentence. "You are a real estate copywriter who writes warm, specific listing copy for buyers." That one sentence sets the voice for everything after it. Without it you get a polite assistant. With it you get someone who writes for your market.

Input. Give it every fact you have. The address, the square footage, the beds and baths, the three features buyers actually mention, the buyer you expect, the neighborhood, and the objection you hear at every open house. If you leave a gap, AI fills it with something it made up. It won't stop and ask. It guesses. So give it more context than feels necessary.

Constraint. Set the limits. Word count, tone, format, and what NOT to say. "Under 250 words. Warm, not stuffy. No exclamation points. MLS compliant. Fair housing safe. Don't use the word luxury." Without limits, AI drifts to its blandest middle.

Example. Paste in a sample you like, a listing description or an email that got replies, and say "match this voice." This is the step most agents skip, and it makes the biggest jump in quality. AI is far better at imitating than inventing.

The structure doesn't box the AI in. It tells the AI where to focus.

Eight Jobs to Run Through the Brief

Almost everything an agent uses AI for falls into eight jobs. Write one good brief for each, save it, and reuse it.

Listing descriptions. The highest risk if done badly, because buyers and sellers can spot AI copy from across the room. Give it the property data, the buyer, a voice sample and a word limit. The full method is in AI listing descriptions buyers actually read.

Follow-up sequences. Give it the lead's source, timing, what they said and your last touch, and ask for a five-message sequence with spacing. You edit, then load it into your CRM. See AI follow-up that works while agents sleep.

Social content. Captions, video hooks and scripts. Give it the topic, the platform, the format and the limits. See real estate social media, conversations not content for what to do with the replies.

Market summaries. Give it your MLS numbers and your local data, and ask for a short client-facing summary in plain English. You supply the facts. AI packages them. Check every number before it goes out.

Client messages. The "let me think about it" reply, the price reduction conversation, the buyer who lost a multiple-offer situation. Give it the situation and the tone you want. It writes a first draft faster, and often softer, than you would under pressure. Read it out loud before you send it.

Lead qualification scripts. The first call and the discovery questions. AI drafts them. You practice until they sound like you. Speed still comes first. The script is what you use after a fast reply lands the conversation.

Neighborhood guides. Give it your own notes and local facts, and ask for a guide a buyer would actually read. You add the photos and publish it on your own site.

Open house and post-tour follow-up. Thank-you notes and "here's what else fits what you told me" messages, built from what the visitor actually said. The follow-up happens, and it mentions real specifics.

The Three Mistakes That Ruin the Output

Asking instead of briefing. "Can you write me a listing description?" is a question, and AI answers questions politely and generically. "Write a property description from these specs and these limits" is a brief. AI carries out briefs.

Stopping after one try. The first draft is rarely the last. Say "make it shorter." Say "make the second paragraph more personal." Say "replace the word stunning with something specific." Three follow-up turns usually get you further than the first answer ever will.

Skipping the example. One paste of a sample you like cuts the tone drift and the guessing. It costs you ten seconds.

All three are habits. All three can be reversed in a week of practice.

Prompts Are Part of the System, Not the System

A prompt is one step in a larger system. What matters is where the output goes next. A follow-up draft that sits in a Google Doc earns nothing. The same draft loaded into your CRM, scheduled, triggered by what the lead does and routed to your phone when they reply earns a conversation.

So build the brief, then connect it. Keep the saved brief for each job next to the place you use it. Read what to automate and what stays human to decide which jobs AI should run and which ones you keep, and AI CRM setup that agents actually use for where the output should land.

And review everything before it goes out. AI drafts. You are responsible for what gets sent, including MLS rules and fair housing language.

The Bottom Line

For your agents: stop asking AI questions and start giving it briefs. Role, Input, Constraint, Example. Save one brief per job. Go at least three turns before you judge the draft. Then connect the output to your CRM.

For you, the leader: teach it live at the company meeting, build it into onboarding, and ask for one prompt in every one-on-one. A team that teaches this has something to say when an experienced agent asks why they should join. If you want to see how that plays out, read how to recruit experienced real estate agents and why referrals became our best recruiting source.

FAQ

The best ChatGPT prompts for real estate agents are written as a four-part brief: a role, the input facts, constraints and an example to match. Blake Suddath calls it RICE. The prompts that pay off fastest cover listing descriptions, follow-up sequences, social content, market summaries, client messages, lead qualification scripts, neighborhood guides and post-tour follow-up. Saving one brief per job lets agents reuse it instead of starting from a blank box.

Write the prompt as a brief, not a question. Start with one sentence telling the AI who to be, add every fact it needs, set limits on length, tone, format and words to avoid, and paste a sample of the voice you want. If the first draft misses, ask for changes in follow-up turns instead of starting over.

AI writing sounds generic when it is given too little to work with. A one-sentence prompt leaves the AI to fill the gaps with the safest, most common phrasing. Adding specific property or lead details, a clear audience, a list of things not to say and a sample of your own voice moves the output away from the generic middle.

Yes, if the agent supplies the facts, sets MLS and fair housing limits in the prompt and reviews every word before publishing. AI can produce a solid first draft quickly, but the agent is responsible for accuracy and compliance. Skipping the review is the fastest way to publish copy that sounds robotic or makes a claim the property can't support.

Train a team to use AI prompts by teaching the four-part brief live at the weekly company meeting, building it into onboarding and asking for one prompt in each weekly one-on-one. Blake Suddath's model books those one-on-ones for 45 minutes and runs the company meeting on Tuesdays. Having every agent build a real brief for a real task that week works better than a slide deck.

AI prompt training helps a team recruit because experienced agents join teams where they believe they will sell more houses, and a working AI lesson is easy to show. Blake Suddath built his agent systems around what agents told him they needed, and has said, "This is the stuff I would teach and build so I could recruit more agents." Inviting a recruit to sit in on the weekly meeting lets them watch the training before they decide.