I've been thinking about this a lot lately.
With Claude, Codex, Cursor, etc., I can get an app from an idea to something working ridiculously quickly.
And honestly, that's great.
But then comes the part that isn't as fun.
You put it in front of actual users.
Someone does something you didn't think of.
An API returns something weird.
A background job fails.
A user creates an account and the account doesn't get added.
An AI-generated function technically works, but under one specific condition it does the completely wrong thing.
And suddenly you're spending more time figuring out what actually happened than fixing the problem.
That's the bit I think we're going to underestimate with vibe coding.
We're getting much better at creating software.
We're not necessarily getting better at knowing when the software is quietly going wrong.
That's why I've been building Shipsure. The idea is to give AI-built apps another layer of visibility around the stuff that tends to go wrong once they're actually being used.
Not another AI coding agent.
Not another "paste your error into AI" tool.
More like:
Your app broke. Here's exactly what happened.
I'm still figuring out what the product should become, so I'm genuinely curious about this:
What's the most annoying bug you've had in an AI-built app that took way longer to find than it should have?
Not the biggest bug.
The stupid one.
The one where you eventually found it and thought "there is absolutely no reason this took me 3 hours."
Those are the problems I'm interested in.