Claude +Phone + free code = 18,000/month scanning rooms. 11 days to first client. Full breakdown cover

Claude +Phone + free code = 18,000/month scanning rooms. 11 days to first client. Full breakdown

Andrey Superior avatar

Andrey Superior · @andreysuperior · May 8

View original post

Someone just killed the real estate industry. A phone. A free GitHub repo. And a browser tab.

A guy scanned an entire house with his phone. Uploaded it.

Now anyone on Earth can walk through it in a browser tab. No app. No VR. No agent. No appointment.

Click → you're inside. Every room. Every angle. Every shadow. Photoreal.

The technology is called 3D Gaussian Splatting. It's been on GitHub since 2023. It's open source. It's free.

And freelancers are already charging $300-$800 per scan.

One person. One phone. One weekend. A business.

Here's the full breakdown.

What most people missed

Real estate agents have charged 3-6% commission for decades. On a $500,000 home that's $15,000-$30,000. For what? Showing people around rooms.

3D Gaussian Splatting doesn't replace agents entirely. But it removes the one thing that justified that fee for buyers: the physical tour.

Now anyone can tour 50 homes in one evening from their couch. And the scan that makes it possible costs $200 to produce.

The math that breaks the industry:

The window right now: there are 2 million real estate agents in the US. Almost none of them are offering 3D Gaussian Splatting tours. The agents who adopt this first will close deals faster than everyone else. They know it. Most don't know how to get it done. That's your business.

Part 1. The science (why this is different from everything before)

You've seen Matterport. You've seen 360 photos. You've seen virtual tours that look like you're inside a potato.

This is different.

Traditional 3D rendering uses polygons flat surfaces stitched together. It looks fake. It loads slowly. It requires expensive hardware.

Gaussian Splatting uses millions of tiny "splats" points of color and depth reconstructed by AI from ordinary photos. The result isn't a model of a room. It's a reconstruction of reality. It loads on a phone. It looks like you're there.

The three tools that make this work:

Part 2. The business model

There are three ways to make money with this. Pick one to start.

Model A. The freelance scanner

Show up. Scan. Deliver. Charge per project. No subscription. No recurring clients needed.

One scan per day, five days a week = $1,500-$4,000 per week.

Model B. The subscription service

Scan once. Charge every month for hosting and updates.

Model C. The agency model

Stop scanning yourself. Train others to scan. Take 30-40% of every job.

Part 3. How to actually do the scan

This is the part everyone overcomplicates. It's four steps.

The prompt that builds the client page:

Part 4. Getting your first client

Don't send cold emails. Don't build a website first. Don't spend a week preparing.

Do this instead:

Day 1: Scan your own apartment or a friend's place. Get comfortable with the process. Test that the result looks good.

Day 2: Walk into three local businesses a hotel, a real estate agency, an Airbnb management company. Say this:

"I want to show you something that takes 30 seconds to look at. Can I pull it up on your screen?"

Show them a demo scan from luma.ai doesn't have to be their property. Let them interact with it. Watch their reaction.

Then say:

"I can do this for your property. Takes me 20 minutes on-site. You get a link you put on your booking page. First one is free."

That's the entire sales pitch. The technology sells itself. Your job is just to get it in front of the right person.

The prompt that writes your follow-up email:

Part 5. The numbers at scale

Part 6. Mistakes to avoid

Mistake 1: Scanning too fast. Walk at half your normal speed. Overlap every angle twice.

Mistake 2: Bad lighting. Open every curtain. Turn on every lamp. Avoid scanning at night.

Mistake 3: Pitching the technology instead of the outcome. Nobody cares what Gaussian Splatting is. They care that guests book faster and cancel less. Lead with that every time.

Mistake 4: Building a website before getting one client. Your first sale happens in person with a phone demo. The website comes after.

The real insight

The technology existed for two years. Free on GitHub. Open source. Anyone could use it.

Nobody packaged it simply for the person who owns a 12-room hotel in a mid-size city and has no idea what a splat file is.

That gap between technology existing and someone making it accessible to a specific buyer is where every solo business like this is born.

The code is free. Claude connects the dots. You show up with a phone.

That's the entire business.

→ Follow - I post systems like this every week → Like if you'd try this in your city this weekend → Save this - the prompts are worth bookmarking