How to Use AI as a Real Estate Agent

Pay for it — it's two coffees a month

Start here: you should be paying for at least one AI subscription, ideally two. It's around $20 a month each — roughly two coffees — for a tool that can save you hours a week. The free versions are a teaser; the paid versions are where the real capability lives: the more capable models, higher limits, file uploads, and the features that make it genuinely powerful.

Paying does two things. It unlocks the good stuff, and — this matters more than people admit — it makes you actually use it. When you're paying for something, you're motivated to get your money's worth, and that's when you start discovering what it can really do. Weigh $20 a month against what an hour of your time is worth, and it's not a close call.

Which AI should you use?

The two to start with are ChatGPT and Claude, and they're good at different things — knowing which to grab for which job is the skill.

• Claude leans strong for detailed, thorough, build-oriented work — writing, analysis, coding, and building tools or simple websites. Reach for it when you're actually creating something.
• ChatGPT is excellent for fast everyday tasks, brainstorming, and quick content. Reach for it for the day-to-day.

There are others, each with strengths worth knowing. The point isn't loyalty to one tool — it's learning the toolbox and using the right one for the job. That's why having both subscriptions is worth it.

The setting nobody tells you about: custom instructions

Here's the move that separates people who get real value from people who don't: you can tell the AI, permanently, how you want it to behave. Both ChatGPT and Claude have a "custom instructions" or preferences setting that applies to every conversation. Almost no agent knows it exists, and it transforms the tool from a generic assistant into one tuned to you.

By default, AI is a people-pleaser — it agrees with you, pads answers with filler, and tells you what you want to hear. That's useless for making good business decisions. You want an advisor that challenges you, not a cheerleader that flatters you. Custom instructions are how you get that.

How to actually prompt it

The biggest difference between people who get amazing results and people who get mediocre ones isn't the tool — it's how they ask. "Prompting" just means how you talk to it, and the rule is simple: be specific and give context.

Instead of "write a listing description," say: "write a listing description for a 3-bed, 2-bath character home in [neighbourhood], emphasising the renovated kitchen and walkable location, aimed at young families, warm but not cheesy, about 150 words." The more direction you give — what you want, who it's for, the format, the tone, the length — the better what comes back.

And when you're learning something new, a trick: tell it to explain things simply and go one step at a time. Something like "walk me through this like I know nothing, one step at a time, and wait for me before moving on." It turns the AI into a patient tutor for any task, and it's the fastest way to learn everything else.

Stop writing emails — here's the high-value work

Writing an email is the floor, not the ceiling. Here's the genuinely valuable work AI can do for a real estate business:

• Systems and automation: build automated follow-up sequences so leads never go cold, set up an email newsletter to your past clients, and automate the repetitive things you do every week. This is where the real time savings live — building the machine that runs your business.
• Content and marketing at scale: draft and schedule social content, write market-update emails, turn a new listing into a full set of marketing pieces. The agent who "never has time for marketing" can suddenly maintain a consistent presence.
• Data and analysis: make sense of market data, summarise trends for a client update, or think through a pricing or negotiation strategy out loud — "here's the situation, what am I not considering?"
• Building actual tools: simple websites, landing pages, and automations, built by asking the AI to build them with you, step by step. You don't need to be a developer — you need to be willing to ask.

Where it gets powerful: AI that runs on its own

Everything so far still involves you sitting down and asking. The next step — and this is where it stops being a chat box and starts being a business advantage — is setting AI up to work without you sitting there. Two ideas are changing what a single agent can do.

1. Set it up once, and it runs itself

You can build things that run automatically on a schedule — no one pressing a button. Picture a report that goes to every one of your sellers each week, on its own, pulling the latest activity on their listing and their market. You set it up one time; it runs every Monday morning forever. Or follow-up that fires by itself: a new lead comes in, and a thoughtful sequence of touches goes out over the following weeks without you remembering to do it. The work of staying in front of people — the thing every agent knows they should do and most don't — stops depending on your memory and your calendar.

2. AI that does the whole task, not just answers

The newer shift is AI that doesn't just reply to you — it goes and does a multi-step job start to finish. Instead of "write me a description," it's "take this new listing and produce the description, the social posts, the email to my database, and the feature sheet" — and it works through the whole thing. Instead of "what should I know about this neighbourhood," it's "research this area and give me a client-ready summary" — and it goes and does the research. You're handing off the entire task, not just one step of it.

Put those together and you get something real: a single agent running the marketing, follow-up, and reporting output of a small team — automatically. That's not a prediction; agents are doing it now.

If this is the part that made you sit up — good. It's the highest-leverage thing in AI for agents right now, and it's exactly the kind of thing worth talking through with someone who's actually built it. That's what the free course goes deeper on, and it's what I help agents set up one-on-one.

Beyond the general tools: purpose-built for agents

Beyond ChatGPT and Claude, there's a growing set of tools built specifically for real estate that apply AI directly to agent workflows — generating seller reports, automating follow-ups, running CMAs and valuations, packaging market data. Many are surprisingly inexpensive for the time they save. The modern setup isn't one magic tool; it's a small stack: a general AI or two for thinking and building, plus a purpose-built tool for the real estate-specific work. Together they automate a huge amount of the busywork.