I Built an AI System That Finds, Underwrites, and Contacts RV Parks While I Sleep

Most real estate investors spend their mornings doing the same thing: scrolling LoopNet, refreshing Crexi, calling brokers, hoping something good shows up. I used to do that too. Then I built a system that does all of it before I wake up.

This is not a pitch for software. This is what I actually built for my own RV park acquisition pipeline at Pineapple Acquisitions, and why I think the old way of finding deals is already dead for anyone paying attention.

The Problem With Manual Deal Flow

When I started sourcing RV parks in the Southeast, the process looked like every other investor's. Search a listing site. Call a broker. Get on a few mailing lists. Wait for something to come across my desk.

The issue is not that the deals are not out there. It is that the best ones never hit a listing site. The owner has held the park for 20 years, has no broker, and is not actively marketing. The only way to find that deal is to go find them directly, one by one, across state licensing databases, county tax records, and environmental permits.

That takes hundreds of hours. Or it takes a system.

I chose to build the system. And the gap between what I can see now versus what I could see six months ago is staggering. Most investors are competing for the same 20 deals on a listing site. I am looking at hundreds of off-market parks that nobody else is contacting because they do not have the infrastructure to find them.

Where the Deals Come From

The system pulls from two categories of sources, and the distinction matters more than most investors realize.

Off-market sources: the real competitive edge

These are parks that are not listed for sale anywhere. Nobody else is calling them unless they are doing the same kind of systematic direct outreach. The system scrapes:

• State licensing and permit databases across seven states: Florida DBPR, Georgia health inspections, South Carolina DHEC, North Carolina EHIDS, Mississippi water facilities, Alabama environmental reports, Tennessee TDEC
• Google Maps county-by-county crawl across all seven states, pulling every RV park and campground listing
• OpenStreetMap as a geographic coverage fallback
• Property tax records for owner information, assessed values, acreage, and sale history
• Skip tracing services for phone numbers and emails when the public record does not have them

This is the deal flow that most investors never see because they are not willing to build the infrastructure to find it. When you are the only person contacting a park owner who has never listed, you are not competing on price. You are having a conversation nobody else is having.

On-market sources: still valuable, different advantage

Listed deals have asking prices, broker contacts, and sometimes operating financials. That makes initial underwriting faster. The system scrapes:

• LoopNet commercial listings
• Crexi investment property platform
• RVParkStore.com, the largest dedicated RV park marketplace
• ParkBroker.com, a specialized RV and MHP brokerage

On-market deals get the same scoring and outreach as off-market ones. The system tags everything so I can filter by source and market status in a single view. If a park shows up on both LoopNet and my state permit scrape, I know it is listed. If it only appears in a government database, I know I might be the only buyer reaching out.

The Buybox

Not every RV park is worth looking at. The system filters for a specific acquisition profile before anything else happens:

• $500K to $4M estimated value
• 40 to 250 sites
• 70%+ occupancy
• 8%+ cap rate
• Seller financing preferred, with parks held 10+ years automatically flagged as seller finance candidates
• Geography: Florida, Georgia, South Carolina, North Carolina, Tennessee, Alabama, Mississippi

Everything outside the buybox gets filtered out before I ever see it. This is not about casting a wide net. It is about only looking at parks that match the deal structure I am targeting: strong cash flow, tax-advantaged through bonus depreciation and cost segregation, and acquirable with creative seller financing.

The AI Underwriting Layer

This is where most deal-finding systems stop. They find leads and dump them into a spreadsheet. Mine does not.

Every lead that passes the buybox filter goes through an automated underwriting analysis using public data. No spreadsheet. No manual research. An AI underwriting agent processes each deal and produces a full financial summary:

• Estimated NOI based on site count, regional average lot rents, assumed occupancy, and standard RV park operating expense ratios
• Estimated property value at an 8% cap rate, cross-checked against county assessed value
• Seller finance probability based on how long the current owner has held the property and whether there are outstanding liens
• Site density (sites per acre) as an infrastructure quality and expansion potential signal
• Ownership entity type to understand whether I am approaching an individual, a family trust, or an LLC

Each deal gets a score from 0 to 100 with a full breakdown of every factor that contributed to the score. By the time a deal reaches my screen, I already know the estimated NOI, the likely value range, and whether the owner is a good seller financing candidate.

Two Tiers: High-Conviction Calls vs. Automated Outreach

Not every qualified lead deserves my time on the phone. The system splits everything into two tiers based on the underwriting score and contactability:

High-conviction calls (Tier 1). Score of 55 or higher AND the system found a verified phone number. These are the parks that are the right size, in the right price range, with a reachable owner who has likely held the property long enough to consider seller financing. A call task gets created and assigned to me with the full underwriting summary attached.

Automated outreach (Tier 2). Everything else that passed the buybox. These get a fully automated outreach sequence with no manual effort from me. If they respond, they automatically move to Tier 1 and I get a call task.

This is the critical distinction. I am not cold calling 700 park owners. I am calling the 30 to 50 that the system already identified as the highest-probability deals based on real financial data. Every other qualified lead still gets professional, personalized outreach. I just do not spend my time on them until they signal interest.

The Outreach Sequence

Every qualified lead gets touched automatically through a multi-channel sequence:

• Day 0: Ringless voicemail drop with a personal message from me, referencing their specific park
• Day 1: Personalized email with property-specific details from the underwriting analysis
• Day 7: Follow-up email
• Day 14: Final check-in email

If a lead responds at any point through any channel, the system moves them into an active conversation stage and I take over personally. The voicemail script is short, warm, and direct. No pitch. No pressure. Just a real person reaching out about their park.

Meanwhile, the AI underwriting agent continues running in the background. As new public data becomes available (a tax reassessment, a sale on a neighboring parcel, a new permit filing), the underwriting summary updates and the deal score adjusts. The system is not static. It learns.

What I Actually See Every Day

When I open my CRM in the morning, I see a filtered view of high-conviction deals. Each one shows a name, location, estimated value, deal score, full underwriting summary, and contact information. I can filter by:

• State
• On-market vs off-market
• Source (which database or listing platform the lead came from)
• Score
• Seller finance likelihood

I do not search for deals. I do not underwrite deals manually. I do not send voicemails or emails manually. I do not review hundreds of leads to find the handful worth calling.

The system does all of that. I pick up the phone and have real conversations with park owners who are already in my buybox, already underwritten, and already contacted by my outreach sequence.

The Math That Makes This Worth Building

One RV park at $3M with 50% cost segregation produces roughly $275K in tax savings plus $120K to $144K in annual cash flow. That is $400,000 or more in year-one economic value from a single deal.

No amount of manual searching, cold calling, or broker relationships produces that outcome as reliably as a system that finds, underwrites, scores, and contacts every qualified park in seven states every single day.

The infrastructure took time to build. But now it runs. And the compounding advantage of having a machine that works while I sleep is something a spreadsheet and a phone will never match.

What This Means for Investors

If you are an investor still sourcing deals manually, I want you to think about two things.

First, the information asymmetry. The investor with a system like this sees every park in seven states, updated daily, scored and underwritten. The investor without it sees whatever a broker sends them or whatever shows up on a listing site. That is a structural disadvantage that compounds over time.

Second, the time leverage. Every hour I used to spend searching, researching, and cold calling is now spent on conversations with pre-qualified owners. The system handles everything upstream. My time goes to the highest-value activity: building relationships and structuring deals.

The gap between operators who build systems and operators who rely on manual effort is already wide. And it is getting wider every day.

If you want to talk about how AI systems apply to your own deal flow or business operations, book a strategy call or read more about why I am buying RV parks in 2026.