I'm Using Agentic AI to Close More Deals (Here's My System)

TLDR
The source calls agentic AI the biggest opportunity of our lifetimes and uses it for underwriting, comps, skip tracing, analytics, mailing-list prep, and CRM upkeep. It does not replace your judgment; you still set the buy box. It can take over manual tasks that used to consume much of the day.

Table of Contents

The Four Controls of a Deal

I’ve been investing in real estate for over 15 years. 150 doors, $2 million a year in rent revenues, 300+ flips. For the last few weeks I haven’t been doing much except working with agentic ai, because the technology took a real turn for the better over the last month or two.

Once I caught wind of this, it became obvious to me that everyone needs to be on the ship. My blunt line in the recording was, “If anybody tells you any different, they’re idiots.” I saw tools that could do repetitive work consistently and reduce labor and expense.

I said I was spending about 100 hours a week on it at the time. That shows my level of obsession, not a setup-time requirement or a promise about how many hours another investor should spend.

Every real estate deal has four controls:

  1. Getting the deal, meaning buying the house.
  2. The strategy, meaning what you do with it to monetize.
  3. The work, meaning managing the subcontractors and the project.
  4. The market, meaning selling it or renting it.

Today I’m only talking about the first one, because it’s the most helpful part and it’s far and away the most important. Get a great deal on the front end and everything else gets easier.

Step One: Define the Buy Box Yourself

Agentic AI is not going to help you with this. You need your brain.

The thing about AI is most people try to make it their brain. It isn’t. It’s a tool you program with the things you’ve already decided. You have to think about the foundation. You have to decide what you want. Then you have to tell it.

So you define your own buy box first. I’ve made plenty of videos on the nine steps I use to build a buy box. Once you have it, everything downstream gets filtered through it.

Then you market however you’re going to market. Real estate agents, wholesalers, or direct marketing yourself. Sending mail, running cold calls, whatever. But before you do any of that, you want the buy box locked in so you can say, “This is exactly what I want to buy. This is exactly what I want to spend marketing dollars finding.”

Key Concept
AI without a buy box is noise. The buy box is the thing that makes every downstream workflow useful. Without one, the AI can give you a pile of leads that look impressive and almost none you would actually buy.

The Full Underwriting Skill

Once leads are coming in, this is where AI gets really helpful. I’ve built a handful of skills, which are workflows the AI can run end-to-end, that I call the full underwriting.

The four pieces: underwriter, comps, skips, and analytics.

Comps

I send the agent a property address. It already knows I’m in Chattanooga and what I consider a comp because those standards are loaded into the agent’s instructions.

It goes out, finds houses that fit my comp structure in my neighborhood, and pulls them back. Then it uses price per square foot to tell me what the house will be worth fixed up. Seconds, not hours. Right away I have my comps and I know what the house should sell for.

Skip

The skip agent finds owner and portfolio context for a property. In practice, that data should be limited to what the operator is authorized to access and use for the specific workflow.

Now I can get on the phone with them. If I was driving for dollars and saw the house, I’ve got the phone number in my pocket. If I already had them on the phone, I have all the context about the rest of their portfolio.

Analytics

This one pulls assessor and register-of-deeds records, including tax value, property facts, and ownership history, then flags items for a human to review.

If you’ve seen my other videos, I talk about DOA, which stands for dead on arrival, as the list of things that kill a deal before you even negotiate. The analytics agent looks for those specifically:

  • Zoning issues
  • Easements
  • Weird items in the title or ownership history that need review

Now I know within seconds whether to pursue the property or pass. It might just suck, and I’ll give a lowball offer. Or there’s a title issue and I skip it entirely. Either way I save a lot of time and avoid deep-title surprises.

The value in the source is that this research can happen before Ross goes deep into a deal or spends money in title. The agent can flag an issue quickly, but a fast flag is not the same as a title opinion, survey, zoning determination, or final underwriting decision.

Marketing and List Management

I send mail, and right now that’s basically my only marketing channel. Send the same list out every month. It works.

But the prep used to kill me. Pull the records. Compile them. De-duplicate. Refine the list. Send it to the mail house. Massive manual work of moving data around on the internet, and that’s exactly the kind of job agentic AI should do.

I walked the system through the steps I use each month. Now it can carry out that process for me.

“Pull the records, run the process, prep it.” I don’t trust the output 100%. I still review before it goes out. But the labor is gone. That gives me hours back to do the things that actually require me.

The list itself, I’d never let AI build. How would it know what I know after 15 years? Maybe over time it learns some of that. But right now my list is based on my experience, my architecture. That’s the point: you set the architecture, AI runs it.

The Full-Beard CRM

I use Airtable as my CRM. I had the AI build it out exactly how I wanted.

Before, my CRM always looked like a guy trying to grow a beard. Spotty, patchy, some fields filled, some blank, some from six months ago. I’d start columns with good intentions and never upkeep them. Then I’d be like, “I don’t feel like building this out all the way.”

Now the AI fills every column every time. Rule. Full beard.

Ross talks to the system through Discord and describes having it use call information to update the CRM. That workflow assumes the operator has lawful access to the recording or transcript and limits access to the people and systems that need it.

The CRM gets upkept by the AI. I make the decisions. I go out and have the conversations. I do the sales. The AI handles the data hygiene that humans always skip.

What AI Will Not Do For You

Here’s where I draw the line.

AI won’t know what you know after 15 years of doing this. Not yet, and maybe not ever for your specific edge. Your buy box, your list criteria, your comp structure, your deal standards. Those are yours. You feed them in.

AI shouldn’t be the one making the sales calls. Don’t send out mass AI texts. Don’t use AI phone calls to sellers. That’s not what this is about. The whole point is: get the best leads in front of you, then you go have the conversations. You do the selling. AI does the moving of data on the internet.

Common Mistake
A lot of people try to turn AI into the salesperson. That’s a step backward. The value isn’t that AI replaces you in front of the seller. The value is that AI gives you more high-quality leads to stand in front of.

Companies are trying to black-box all this stuff. Easy to use, no idea how it works. One path is to just use the tools and ride the improvements. The other is to learn the underlying mechanics. Both paths work, but if you want to be techie about it, you need to put in the hours. If you watch YouTube videos on agentic AI, a lot of the creators are already old-school. They’ve been around since the start and they built up knowledge layer by layer. You’ll feel like you’ll never catch up.

You won’t feel that way in a year. Right now is the ground floor. This is the opportunity to understand how the biggest skill of our careers is put together, right from the start. Don’t sleep on it.

FAQ

I’m not technical. Can I still use this?

The source describes two paths: use easier tools that hide the underlying mechanics, or spend more time learning how the systems work. It does not name a product or say that every workflow can be built without technical help.

How do I set up my first agentic workflow?

The source’s pattern is to define the human judgment first, then show the system the manual steps and review its output. It does not give a fixed setup recipe or promise success within a set number of attempts.

Should I automate contacting sellers?

No. Automate the research, not the conversation. The research is what eats your day and doesn’t require your voice. The conversation is where you win the deal and it has to be you.

What AI tools do you use specifically?

The recording does not name a specific AI product. It mentions using Airtable for the CRM and Discord as the place where Ross talks to his system. The durable point is the workflow: define the buy box, let the system prepare research and data, and keep sales and final decisions with the operator.

How much time do I need to invest to get started?

The source does not give a setup-time or weekly-maintenance estimate. Start with one bounded workflow and keep human review in place until its outputs are reliable.