我用 Claude Code 造了个"互联网法庭" LoopProof，现在想知道有没有人真的需要它

A couple of weeks ago, I had a domain sitting around:

loopproof.me

I owned the domain, but I didn't have a product.

So I opened Claude Code and gave it a pretty unusual instruction:

“Build anything you want. Here's the domain. It's already registered and it's mine.”

I didn't give it a startup idea.

I didn't give it a detailed product specification.

I just let it come up with something.

It came back with the idea of an Internet Court.

The concept was simple:

Someone has an argument.

They put the argument on trial.

Other people vote FOR or AGAINST.

The votes become the verdict.

I thought it was ridiculous.

Then I thought...

okay, let's actually build it.

And that's how LoopProof started.

What I actually built

The idea became a real production app instead of staying as a Claude conversation.

The basic flow is:

1. Someone files a case explaining their side of an argument.
2. The case gets a public URL.
3. Other people can read it and vote FOR or AGAINST.
4. The vote count updates into a live verdict.
5. The case can be shared with other people.

I also ended up building the less-fun parts that come with making a real product:

Authentication, database logic, reporting, comments, realtime updates, and subscriptions.

The stack ended up being Next.js, React, and Supabase, with PayPal handling the paid membership side.

The thing I cared about most was one rule:

I didn't want to fake the activity.

No seeded votes.

No fake users.

No pretending the site was popular when it wasn't.

If a case has zero votes, it has zero votes.

And that's where things got interesting.

Then I launched it

This is the part where I thought people might actually use it.

I had spent days building.

The website was live.

The cases worked.

Voting worked.

The verdict worked.

So I put it in front of people.

And...

almost nobody came.

I posted it on Reddit.

Some people asked, “What's the point?”

Some people weren't interested.

And honestly, that hurt.

Because when you've spent days building something, it's very easy to think:

“Maybe nobody wants this because the idea is bad.”

But I don't actually know that yet.

Maybe the idea is bad.

Maybe the product isn't solving a strong enough problem.

Maybe I'm explaining it badly.

Maybe people need to see an interesting case before they understand why they would use it.

Maybe I just have no distribution.

That's the problem I'm trying to solve now.

Not another feature.

Users.

What I'm trying to figure out

Right now, I don't think adding more features is the answer.

The product already does the basic thing I wanted it to do.

The harder question is:

Why would someone choose this over their group chat, Reddit, Discord, or just arguing with their friends?

I think there might be something interesting about turning an argument into a link.

Instead of:

“Bro, you're wrong.”

and then 40 messages later...

You could have:

Take it to court.

One person writes the case.

Other people vote.

Eventually there's a result.

That's a much more specific use case than “another social network.”

But I'm still testing whether that idea is actually useful to other people, or whether it is only fun to me because I built it.

And that's probably the most uncomfortable part of building something yourself:

You can prove that you can build it without proving that anyone needs it.

So here's the experiment

I decided to keep the project alive instead of quietly abandoning it.

I'm not posting this because I think I've figured everything out.

Quite the opposite.

I'm posting it because I'm trying to answer a question I can't answer by myself:

Would you actually use this?

Imagine you and your friend are arguing about something completely stupid.

Who is wrong?

Instead of arguing for another hour, you could put the argument on trial and let other people vote.

Maybe that's useful.

Maybe it's just a gimmick.

I'd genuinely like to know which one you think it is.

What I'm taking away from this

I started this project thinking the hardest part would be building the product.

It wasn't.

Claude Code helped me get from an empty domain to a working application much faster than I expected.

The difficult part came afterwards.

Once something is live, there is no prompt that can guarantee people will care about it.

You have to find out whether the problem matters, whether the idea is understandable, and whether people have a reason to come back.

I think that's the part of building with AI that gets talked about much less.

The barrier to making software is getting lower.

The barrier to making something people actually want hasn't disappeared.

I'm still figuring that part out.

And honestly, I'm glad I built it even though the first launch didn't go the way I hoped.

At least now I have something real to learn from.

AI disclosure: This article was written with the assistance of AI. Claude Code was used extensively during the development of the project, and ChatGPT assisted me with drafting and editing this article. The experiences and project described here are my own.