I Built a ChatGPT-Powered Real Estate Deal Finder

Intro: My Morning Email from ChatGPT

Every morning, I wake up to a personalized email from my new favorite virtual assistant—ChatGPT.

This daily email is a curated, data-rich summary of the best potential rental properties currently listed on the MLS in my local market. It includes everything I need to make fast, informed decisions. We’re talking addresses, listing prices, days on market… and it goes even deeper than that. This email includes a first-pass at underwriting each property: this includes estimated rents, after repair values (or ARVs), cap rates, and even projected monthly cash flow after all my expenses.

And the best part? I don’t have to lift a finger. It shows up before I even roll out of bed (… thank you, ChatGPT).

Today, I’m going to walk you through exactly how I built this system using just four no-code tools. That means no programming experience required. Whether you’re a new investor or already managing a large portfolio, this system can save you hours of work every week and more importantly, surface great deals you might otherwise miss.

So let’s dive into how it all works—and how you can build your own version in just a single afternoon.

The Tech Stack: Simple, Automated, and No Code

This system runs on just four tools. Each one plays a specific role, and together they create a seamless, automated machine that works while I sleep.

First up is Make.com. Think of Make as both the glue and the brain. It handles all the automation logic—scheduling the daily run, triggering the API calls, organizing the data flow, and ultimately sending me the final email. Everything is visual too—no code, no scripts. You just drag and drop your logic blocks, connect your tools, and you’re off and running.

The second tool I use is RentCast, which powers the entire system with data. Through their API, I get access to a national database of on-market listings, property information, rental estimates, property value estimates, and even a list of comps. I use RentCast to pull new listings that match my basic buy box. For example, all active on-market listings within a 10-mile radius around my home and matching a specific property type (like single family).

Once that data is pulled, it flows into Airtable—the third tool in my tech stack. Airtable is where the filtering and calculations happen. I apply my more granular buy box criteria—things like no HOA fees, specific bed and bath counts, listing price ranges, and maximum age of the property. I also run initial underwriting right inside Airtable: calculating my net operating income, cap rate, and cash-on-cash return, based on standard assumptions I’ve refined for my specific market.

And finally, the last tool in my tech stack, and perhaps the most powerful, is ChatGPT. This is where the magic happens. ChatGPT takes all the structured property data and turns it into something incredibly readable and actionable. It organizes the top deals into a clean format, identifies any patterns or standout opportunities, and even writes a personalized email that summarizes the most important takeaways. It’s like having your own full-time real estate analyst who works nights, doesn’t complain, and costs next to nothing (... on a side note, am I a bad manager?).

The Tip of the Iceberg: What I See in My Inbox

Let’s talk about what this actually looks like for an end user, such as myself.

You can think of this system like an iceberg. Most of the action happens below the surface, but the part I interact with every day—the tip of the iceberg—is what really matters to me. It’s what makes this system usable, valuable, and frankly fun.

Each morning, I open my inbox and see a breakdown of the top 15 on-market rental deals that fit my criteria. Let me walk you through a real example.

For every property, I’ll get a bulleted list of over 14 metrics that you can see in this example.

A lot of information right? But it's all organized in a way that’s pretty glanceable and it avoids me having to go back looking for specific information.

So in addition to these metrics, ChatGPT gives me a human-style summary. It highlights how many total properties are in today’s list, how many of those are new vs. repeats from prior lists, how many have dropped in price, and how many meet my must-offer criteria— in this case, producing net cash flow over $200/month or an 8% cap rate.

That kind of commentary gives me the big picture at a glance—so I know whether to dig in deeper or move on with my day.

This daily report is more than just data—it’s a decision-making tool. It keeps me consistent, saves me time, and gives me confidence that I’m not missing hidden gems in the market.

Below the Iceberg: How the Workflow Actually Works

Now, let’s go below the surface of this iceberg—and discuss the backend of how this whole thing actually works. As a full disclaimer, we’ll be getting a little in the weeds here so no hard feelings if you want to skip ahead to the next section of this video.

We’ll start with Make.com. Every morning, at 4:30am my local time, Make.com kicks off the workflow with an automated HTTP request to RentCast’s API, pulling roughly 500 listings that match my base-level buy box. These are active single-family on-market properties within a 10 mile radius of my house.

From there, Make filters down to the 100 cheapest listings based on list price and sends them into Airtable.

Airtable then goes to work applying deeper filtering—removing any properties that don’t match my must-haves: such as no HOA fees, a minimum bed/bath count, a listing price range, and a maximum age of home.

Once that refined list is created, this could be for example around 30 properties, the list is sent out of Airtable and back into Make.com, where it’s filtered again based on the 15 cheapest properties based on price per square foot.

This is our final top 15 property list for the day.

At this point, Make.com enriches the information we have on these properties by requesting additional data from RentCast—specifically, pulling rent estimates, property value estimates, and comps.

All of that data is stored back in Airtable, which then performs a full underwriting analysis using my own custom assumptions on OPEX, CAPEX, and financing. These assumptions are based on my local market knowledge and can be easily tweaked as rates or rehab costs change.

For example, I often estimate the cost of repairs and maintenance, capex, and vacancies as 5% of rent each, or 15% of rent total. I also include assumptions for property taxes, insurance, utilities, property management fees, and financing costs.

Once everything is fully underwritten by the application, the top 15 properties are sorted by their estimated net cash flow and passed back into ChatGPT.

ChatGPT analyzes the properties, compares today’s list to previous days, highlights changes like price drops or returning properties, and writes a clean, professional email that includes both the data and the analyst commentary.

Make.com then sends the email directly to me, and stores a copy in Airtable so I can analyze trends and historical analyses in the future.

Yes—it sounds like a lot. But the beauty is, once it’s set up, it runs completely on autopilot. All while I sleep, and I never have to touch it.

If you’d like a deeper dive into any part of this system—RentCast API calls, Airtable formulas, ChatGPT prompts—let me know in the comments. I’d love to do a more technical follow-up newsletter.

Also, I’m considering offering this as a done-for-you service for other real estate investors. If you’d like to be a beta tester and get daily emails of the best on-market deals in your market, based on your specific buy box, without having to build anything yourself—simply email me at ben@reiplusai.com. It would be great to work together.

Conclusion: This Is Just the Beginning

So in summary, I built a ChatGPT-powered real estate deal finder that sends me a daily, curated list of the top MLS rental properties in my market—already filtered and underwritten to match my investing strategy.

But this is just version one.

Every week, I’ll be improving this system. Adding new data sources. Refining the logic. Exploring how it can be adapted for different strategies—whether that’s flipping, BRRRRs, or short-term rentals.

I’ve got some ideas. For example, I could build a flip-focused version of this that prioritizes low price-per-square-foot, big ARV spreads, and long days on market. Or a short-term rental version of this using data from AirDNA or Mashvisor to find listings with high Airbnb potential.

The possibilities are endless and I’m excited to continue digging into this and documenting progress right here on this newsletter.

If this helped you—or sparked any ideas—please subscribe. As of right now, I have exactly two subscribers, and one of them might be my mom... So if you want to see more real estate meets AI content, this is your moment to help me, help you.

Thanks for reading, and I encourage you to check out another one of my recent newsletters linked below.

P.S. Did you find value in this newsletter? If so, please forward this email to somebody you know. You're not only helping them (by giving value), but you are SIGNIFICANTLY helping the REI + AI channel.

Thank you for the support!

SUBSCRIBE TO THE NEWSLETTER

This Week's Video