Build guide

How I Use Claude to Find Cash Flow Positive Rentals

Underwrite rental listings with plain Claude in the browser: screenshots in, identical cash-flow analysis out on every property.

Claude
May 11, 2026
Optimized for

What you'll build

A system for underwriting rental listings with plain Claude in the browser - no Claude Code, no scraping, no APIs. You feed it full-page screenshots of Realtor.com listings, and it runs identical cash-flow analysis on every property, filters out the ones that don't pencil, and hands back a printed-ready PDF: a one-page summary per property, an optimized tour route, a tour-day prep checklist, and questions to ask every listing agent.

I built this when my dad decided to look at rental property in New Britain, CT - an hour-and-a-half drive away - with a realtor family friend. I started with 40+ listings from my saved searches and walked into every showing already knowing the numbers. Even the realtor, 30-40 years in the game, works off the dome and a 2-minute Zillow skim; this gets you the in-depth research off the rip.

This is the follow-up to my deal-hunting video (the saved-search + scheduled-task setup): https://www.youtube.com/watch?v=MtlE14HC9G0

What you'll need

Step 1: Run your saved searches on Realtor.com

Open Realtor.com and run the saved searches for your area. Open each results page (or listing) in its own tab. Mine surfaced 40+ properties across the individual searches.

Step 2: Capture each page as one full-size screenshot

Realtor.com has no Claude integration and blocks scraper/Playwright-style automation. Don't fight it - bypass it. Claude reads the pixels instead.

On each tab:

Right-click → Inspect   (or Cmd+Option+I)
Cmd+Shift+P
Type: "full size screenshot" → Capture full size screenshot

Chrome saves one tall image of the entire page - every detail, no stitching multiple screenshots together.

Step 3: Paste the screenshots into Claude with the cash flow prompt

Dump the screenshots into Claude. My prompt was nothing complicated - close to this, verbatim from the video:

I'm sending a bunch of screenshots of properties in New Britain. You're going
to run an analysis on them in terms of cash flow - whether they're cash flow
positive, number of beds and baths, the cap rate. The ones that don't make
sense, disregard. The ones that do make sense, aggregate them all - I'm going
to book tour dates for them. We're okay with a 20% down payment.

Claude will come back with questions about how to format the response - answer them. (I could make this prompt 10x better just by asking Claude to improve it, but simple worked.)

The consistency is the point: a human analyzing property #16 is not as sharp as they were on property #1. Claude runs the identical analysis at the same level on every single one.

Step 4: Let it filter and aggregate

Out of the 40+, Claude kept 11 strong (or relatively strong) candidates and disregarded the rest. The output became a ~12-page PDF, one page per property:

Step 5: Build the tour route and itinerary

Since I was driving 90 minutes, I had Claude aggregate all the shortlisted addresses and map the most efficient route. It also produced a tour itinerary: top priorities, which agent to contact for each property, their numbers.

Step 6: Print it and carry it

I printed the whole packet and had it in hand at every showing. On the drive up I read each property's page, so when a listing agent raised something, I already knew it. The parts that earned their keep:

That's it. None of this replaces a pro - but it means you're not walking in blind. And it cuts both ways: a real estate agent could run this standing search for a client and send them polished reports automatically, and a buyer can do their own due diligence before ever getting in the car.

Resources

Some links in this guide are affiliate links. They cost you nothing and support the free guides.

← Back to the library Good luck and happy building