I’ve been building BeHooked for two years. I’m a self-taught dev and I’ve been learning to code on my own the whole time.

But I had never gone all in on vibe coding.

I had tried a few things with Claude Code and similar tools. Small experiments. Nothing serious. I always stayed on the business side and let someone else touch the code.

This time I wanted to actually do it.

A free Lovable subscription

It started with LinkedIn Premium. It came with a free Lovable subscription, so I thought, fine, let me see what this thing can do.

To test it, I built a link-in-bio tool. Basically a clone of Bento.me.

It came out looking promising. Nice, even. I wanted to use it for my own profile.

There was just one problem. I wasn’t ready to pay Lovable every single month for it.

So I asked myself: okay, what can I do?

Getting off Lovable

I pushed the whole thing to GitHub and made sure I could work on it with Claude Code.

First job: strip out everything Lovable-specific from the codebase.

Second job: figure out how to host it for free.

Cloudflare was the obvious answer, so that’s what I used. Supabase is also free, so that became the database.

It was great. I had my own tool, on my own infra, costing me nothing.

I was doing all of this over weekends.

A free Codex subscription

Then I went to an event called the World’s Biggest Hermes Hackathon, and I walked out with a $200 Codex subscription for free.

That was the push I needed to finish the whole thing properly.

The link-in-bio tool itself was done within a day or so. But then I hit the real question.

Who is going to pay for a link-in-bio tool? You get those for free.

I wanted to solve an actual problem.

The actual problem

While researching, I got really inspired by Stan Store and Linktree.

I noticed all of them had an Instagram auto-DM feature. And all of them worked the same way: you go to Instagram, you post, then you come back to their platform and set up the automation.

It’s so hectic. Who is going to do that every time?

It would be so much better to just ask ChatGPT or Claude, connected to an MCP, to “upload this video to Instagram and set up the automation.” One instruction. Done.

That became the problem I was building for. And that’s the moat I’m building towards.

A Cursor subscription

My Codex subscription ran out after a month or so. By then only the last few parts of the app were left before launch.

Two weeks later I went to another event, a Cursor one, and got Cursor credits there. I bought the Cursor Pro plan for $20 and finished the whole project on that.

Where it stands

The project is Bento. Links, products, a store, a post scheduler and DM automation, all in one place.

It’s not launched yet as of writing this. But it will be soon.

Free Lovable to test the idea. Claude Code to make it mine. Free Codex to build the real thing. Cursor to finish it.

Not a bad way to build your first indie hacking project.