I bought comidavenezolana.com over a decade ago. It just sat there. I kept paying the renewal every year but never actually did anything with it. The idea was always in my head: a marketplace connecting Venezuelan food vendors, a directory of restaurants, a way to help my community sell what they make. But building it felt impossible.
The old way
Back then, if you wanted to create something like this, you had two paths, and both were painful.
You could buy a WordPress template, install it yourself, host it somewhere, and manually add restaurants one by one. Or you could spend weeks scraping data, cleaning CSVs, and uploading them into some clunky admin panel. Even with automation, you were looking at a lot of manual work.
As a nontechnical person without a CS background, the barrier felt real. The ROI wasn’t clear enough to justify months of grinding. So the domain just kept renewing every year, a reminder of an idea I couldn’t execute.
The pivot
A few months ago, I started building the original vision: a marketplace for people making Venezuelan food at home. Hallacas in December. Pabellón on a Tuesday. Real money in a cultural moment.
Then I hit the wall. Permits, liability, state regulations. Suddenly I wasn’t just building a marketplace, I was potentially liable for facilitating commerce that wasn’t fully legal in every jurisdiction. That’s not a product problem, that’s a legal problem.
But the idea itself was solid. People want Venezuelan food. The demand is real.
What if I flipped it?
The realization
Instead of home cooks, what if I aggregated restaurants? Real businesses with real permits. Pull their data from Google’s API, layer in reviews from multiple sources, and create a fresh directory that ranks on SEO. Then, once I have traffic and authority, monetize. Restaurants pay to be featured prominently in their region. Doral, Florida. Search for arepas. There they are.
That’s not just a marketplace. That’s discovery infrastructure. And it’s defensible.
And most importantly, it’s legal. No liability. No gray area.
The build
This is where the math flips completely.
I started building on a Monday. By Wednesday morning, I had a working prototype. The stack:
- Astro for the front end
- Claude Code for everything else
- Supabase for the database
- Vercel for deployment
- Google Places API pulling live restaurant data
- Review aggregation from Google and Yelp
A clean, fast directory that actually works.
The old way? WordPress template, manual setup, CSV uploads, days of configuration. Maybe a week of solid work just to get the foundations in place.
The new way? Claude Code. Tell it what you need. Let it build. Iterate. Ship.
A day and a half.
What changed
It’s not that this was impossible before. You could always do it. But the friction was real. You needed to know WordPress, or hire someone who did. You needed to manage servers, handle CSVs, do manual data entry. All of that consumed time and energy better spent on the actual business.
Now? Even a nontechnical person can ship an MVP in a day. You can validate whether the idea works before you commit months to building it perfectly.
That’s the actual shift AI brings. It’s removing the friction between idea and execution.
The monetization layer
Once you have a quality directory with good SEO and fresh data, monetization becomes obvious. Restaurants want to be found. They want to be at the top of search results in their region. They’ll pay for that.
But there’s more. Onbranded, my platform for restaurant operators, becomes a natural upsell. Restaurants claim their profiles, then see the opportunity to automate their operations, manage reviews across platforms, and build better websites. The directory isn’t just a standalone product, it’s a distribution channel.
And for restaurants without decent websites, I can offer that too. A custom landing page, verified listings, real photos. A proper web presence for businesses that might never have had one.
The broader point
I’ve watched technology waves come and go: ecommerce, Amazon FBA, mobile apps. I usually see them early but never go all in. With AI, I’m making a different choice. This time, I’m committing.
Not because I suddenly became a great engineer. I’m still not. I don’t write code. I describe what I want and Claude Code builds it. But that’s powerful enough.
The domain that sat dormant for over a decade? It’s alive now. An MVP shipped in a day and a half. Traffic coming in from SEO. Restaurants starting to see the value.
Will it work? I don’t know yet. But at least now I get to find out.
And that’s the real win with AI: you don’t have to choose between having the idea and having the time to test it. You get both.
The ice cream guy is building in public. Follow along. 🍦