Your MLS Is a Gold Mine: How Real Estate Agents Can Use AI to Mine It

Jerad Larkin

14 hours ago

10 min read

How do real estate agents use MLS data with AI? Run your search in REcolorado, select your results, and export using the Full Export format, which carries hundreds of columns per property instead of the handful you see on screen. Upload that file to ChatGPT or Claude with the thinking effort turned up, and tell it what you want. That one file can become a CMA, a market report, a pricing analysis, listing presentation material, or a neighborhood deep dive for a buyer.

Get the prompts and skills from this class

Everything I handed out in the room is here. Take what you want.

Price Reduction Angle Generator, works in any AI: https://bit.ly/price-reduction-prompt
Market Activity Report, a Claude skill: https://bit.ly/mls-activity-skill
ShowingTime Activity Report, a Claude skill: https://bit.ly/showingtime-skill
The Sit Tax report I open the class with: https://bit.ly/denver-sit-tax

Grab those now if that is what you came for. The written version of the class is below.

I taught this one live in Denver, and I started it the way I start most of them, by admitting I put almost no effort into the thing everybody wanted a copy of.

The night before, I sent out a market report called the Sit Tax. It looks at the median seller give back from list to contract across the Denver metro counties, and what every extra week on market costs a seller. Agents emailed me about it all morning. I built it by exporting MLS data and talking to an AI for about ten minutes.

That is the point of this class. You already have access to the best real estate data set in Colorado. Most real estate agents are using maybe two percent of it.

One export out of REcolorado, handed to an AI, can become any of these:

A CMA for a seller, built off closed comps you selected
A listing presentation with real numbers behind the pricing slide
Market stats for a subdivision, a zip code, or a price band
An analysis of why a listing is sitting, and what to do about it
A neighborhood deep dive for a buyer who is deciding between two areas

https://youtu.be/YpTAQutTyy0

What this class covers

The export setting in REcolorado that pulls hundreds of columns instead of a handful
Why your thinking effort setting matters more than how clever your prompt is
A live pricing analysis on a Centennial listing that had been sitting 14 days
The prompt I use when a seller needs to hear the price conversation
Letting the Claude Chrome extension run InfoSparks for you instead of clicking through it
The ShowingTime report that hands you the buyer agents working your price band

The Full Export is the part almost nobody uses

Here is the process, and it is the same whether you are building a CMA, a listing presentation, or an argument for a price reduction.

Run your search in REcolorado the way you always do. Set your filters, set your area, set your parameters.
Click Results, then select the records you want. You can select all of them.
Click Export.
This is the step that matters. Under export file format, choose Full Export instead of the default.
Download the CSV.

The default export gives you the columns you see on screen. Full Export gives you hundreds of columns per property, including public remarks, private remarks, and every data point the listing agent entered.

Nobody wants to read that. That is exactly why you hand it to AI.

One practical note. Your MLS plan caps how many records you can export per month, and you will see your remaining allowance right on the export screen. Keep an eye on it. If you run out, a PDF export or even a screenshot still works, because AI can read both.

Turn the thinking effort up before you upload anything

This is the setting that separates a useful analysis from a generic one, and most real estate agents have never touched it.

In ChatGPT there is a thinking effort control under the chat box. Drag it to high. In Claude, you pick the model at the top of the window and the higher ones reason harder. Either way, the answer takes longer and it digs a lot deeper.

For a quick question, leave it alone. For a pricing analysis on somebody's largest asset, turn it all the way up. And yes, this is a paid feature on both platforms. Twenty dollars a month. If you are doing real data analysis on the free plan, you are handicapping yourself for no reason.

Then you just talk to it. I do not write elaborate prompts anymore. I upload the file and say something like, here is everything that closed in Cherry Creek in the last 60 days out of REcolorado, analyze this and give me the key findings. The models have gotten smart enough to figure out what I want.

Garbage in, garbage out still applies. The more context you give it about the property and the situation, the better the output. But you do not need to be a prompt engineer.

A live example: a Centennial listing sitting at 14 days

Somebody in the room had a listing at 16575 E Prentice Ave that was not moving. We ran it in front of everybody. Two steps.

Step one, pull the comps

Back into REcolorado. Single family, one mile radius around the subject, closed in the last 60 days. That gave us 39 properties. Select all, export, Full Export, download.

In a perfect world you pick the comps yourself, because you are the expert on that neighborhood. For the sake of class time we let AI pick from the pool.

Step two, pull the subject property

This is the fastest trick in the class and it takes about four seconds. Instead of typing out everything you know about the property, pull up your subject listing in the MLS, click Print, and choose Full Property View, all tabs, print to PDF.

You get a six to ten page document with every field and every photo. Upload that alongside the comps file. AI can read the photos too, which means it can tell you something about condition and staging, not just numbers.

Now the AI has the subject property and about 40 closed comps around it. That is a real pricing conversation waiting to happen.

The prompt I use for a listing that is not selling

This is the Price Reduction Angle Generator, and you can grab it here: https://bit.ly/price-reduction-prompt

What it does: when a listing is not selling, it figures out why based on actual MLS data, and gives you one clear, persuasive angle to take to your seller. It asks you for the property details, analyzes the comps you upload, and delivers talking points rooted in what buyers are responding to right now.

The output is a data backed narrative that makes your price adjustment feel strategic instead of desperate. That is the whole difference between a seller who feels managed and a seller who feels sold.

It is just a set of instructions. Copy it, paste it into whichever AI you use, upload the two files, and let it run.

When the output looks right but reads boring

A spreadsheet analysis is not something you send a seller. It is accurate and it is dull.

Two fixes. The first is to just ask. I will say, make this much more attractive and turn it into a PDF, brand it with the Chicago Title colors, figure it out. It usually does.

The second is Gamma, at https://gamma.app. For a long stretch this was my answer when the AI output looked flat. Copy the analysis, paste it into a new Gamma, tell it to make a PDF document, pick your theme and your colors, and it builds you a ten page report that actually looks like something. It starts free.

If you are at the stage where the AI is getting the numbers right but the documents look like homework, that is what Gamma is for.

InfoSparks, and letting AI run it for you

Inside REcolorado you have InfoSparks, and inside ShowingTime you have a version of it too. It builds market stats by area, price band, property type, bedroom count, all of it.

I do not love the reports it produces on its own, so I had AI build my own version. Same data, my colors, one page, client ready.

Here is where it gets fun. With the Claude Chrome extension installed, I do not go click through InfoSparks at all. I tell Claude the property, it asks me a few intake questions, price range, property type, beds, baths, square footage, and then it opens InfoSparks, applies the filters, pulls the numbers, and hands me the PDF.

We did that live on a Fort Morgan listing at 411 Ensign St. It logged in, ran it, and built the report while the room watched.

A question came up that is worth answering. No, it is not an API connection. There is a real MLS API and it runs around a thousand dollars a month. All I did was stay logged in. The extension works the screen the same way you would.

How to install the Claude Chrome extension

Five steps. Somebody asked, so here it is in order.

Sign up for Claude at https://claude.ai. It is twenty dollars a month for the plan you want.
Install Google Chrome if you do not have it. You do not have to browse in it, you just need it on the machine.
Open Chrome and search for the Claude Chrome extension. It is the first result: https://chromewebstore.google.com/detail/claude/fcoeoabgfenejglbffodgkkbkcdhcgfn
Click Add to Chrome and sign into Claude.
Click the puzzle piece icon in the top right of Chrome, pin the extension, and open the side panel.

The first time you open it you will click through a few permission screens. Read them. This thing can go anywhere on the internet and do things on your behalf, so you should know what you are agreeing to.

After that, you talk to it in the sidebar and it works the page for you.

How do you know the AI got the numbers right

Best question of the class, and the honest answer is that you check.

I went through the Fort Morgan report line by line before I showed it. AI is not perfect and it never has been. What it is, is fast and tireless on the arithmetic and the formatting.

You are the expert on the subject property and the neighborhood. The model is not, and it will not be for a while. The workflow that works is the one where AI does the heavy lifting and you make the call. Keep it contained to the MLS data you fed it rather than letting it roam the internet pulling numbers from who knows where, and you will trust the output a lot more.

The ShowingTime report almost nobody opens

I saved this for the end because it is the single most immediate thing on this list. If you have a listing sitting right now, you could run this today.

Log into the new ShowingTime at https://www.showingtime.com, go to Reports on the left, and look for the Target Market Analysis Report. If you do not have it, it is a phone call to ShowingTime and somewhere around ten or fifteen dollars a month.

What it does: it shows you which price ranges are getting the most showings in an area, broken into price increments you control. Three ways to use it.

One, find the buyer agents and call them

Run the report for your listing area and price band over the last 60 days. Where there is enough data, you can click into a price bracket and it gives you the buyer agents who showed in it, with their phone number and email address.

Now you have a call list of agents with buyers who were shopping your exact price point in your exact neighborhood. Ask if their buyer is still looking. Ask if they would consider showing your listing. That is a straightforward afternoon of work that gets your seller real exposure.

Two, confirm the price with showing data

This is what the report was built for. Adjust the price increments and look at where showings are actually clustering. Sometimes there is a visible gap right above where you are priced, and that is a much easier conversation to have with a seller than a hunch.

Three, run it before the listing appointment

If you are going after a listing you have not won yet, pull this report first. Pull the buyer agents working that area and price range, call them, and walk into the appointment already knowing there are buyers out there for it. That is a different presentation than the one everybody else is giving.

Where Chicago Title fits

I am an Account Executive with Chicago Title of Colorado, and I am in front of Denver metro real estate agents all day. This class came straight out of those conversations, because the same thing keeps coming up: listings are sitting, sellers are asking what else can be done, and the market is slow is not an answer.

My job is to hand you the data, the marketing, and the systems that help you win listings and move the ones you have, and then make the closing smooth when it comes together. The classes are part of that, not a side project.

FAQ

How do I export MLS data from REcolorado for AI analysis?

Run your search, click Results, select the records you want, then click Export. Under export file format choose Full Export rather than the default. That gives you hundreds of columns per property in a CSV, including public and private remarks. Watch your monthly export allowance, which is shown right on that screen.

Which AI is best for analyzing MLS data?

ChatGPT and Claude both do this well. The bigger variable is the thinking effort or model level you choose, not the brand. Turn it up for anything involving numbers. I use Claude because of the Chrome extension, but plenty of real estate agents I work with get excellent results out of ChatGPT.

Can AI build a CMA from MLS data?

It can build the analysis. Export your closed comps, print the subject property as a Full Property View PDF, upload both, and ask for the pricing analysis. You still pick or verify the comps and you still make the pricing recommendation, because you are the licensed party and you know the neighborhood.

What is the Target Market Analysis Report in ShowingTime?

It is a report that shows which price ranges are getting the most showings in a given area, and lets you pull the buyer agents who showed in each price bracket, with contact information. Not every agent has it turned on. It runs roughly ten to fifteen dollars a month through ShowingTime.

Do I need to pay for ChatGPT or Claude to do this?

For serious data analysis, yes. The paid plans are around twenty dollars a month and they unlock the higher thinking levels, bigger file uploads, and the agentic features like the Claude Chrome extension. The free versions are fine for learning, but they will not give you the analysis quality you want on a pricing conversation.

Take one thing

That is the whole ask. Export one search with Full Export and run one analysis on it, whether that is a CMA, a market report, or an argument for a price reduction. Or install the Chrome extension tonight. Or open ShowingTime and see whether you already have the Target Market Analysis Report sitting there unused.

Pick one. The rest of this will still be here.

If you have a listing sitting right now and you want an analysis run on it, or you want marketing ideas to get it in front of the right people, reach out to me.

Want more tools, tactics, and resources like this? Subscribe to my monthly market emails at https://www.milehightitleguy.com. I share real estate marketing ideas, AI tools, and exclusive invites to upcoming classes and events across Colorado.

Every class and event I am teaching: https://bit.ly/mile-high-classes