A few weeks ago I had one of those dumb founder moments.
I was happily using Claude and other LLM tools to build faster, ship faster, and tell myself it was all worth it because I was being "productive".
Then I looked at the bill.
It was not catastrophic. It was worse than that. It was the kind of number that makes you feel mildly stupid because you know exactly how it happened.
A hundred little prompts. Background usage. Testing loops. Switching between tools. "Just one more run." It adds up fast when you are building every day.
What annoyed me most was not even the cost. It was the invisibility.
I could see CPU usage. Memory usage. Battery usage. Network usage. But the thing I was suddenly paying real money for, tokens, was basically invisible while I worked.
So I built TokenBar.
TokenBar is a tiny macOS menu bar app that tracks AI token usage live while you work.
That is the whole pitch.
No dashboards I have to remember to open. No waiting until the end of the month to find out I got reckless. Just a live number in the menu bar so I can catch bad habits early.
The funny part is I originally built it for myself because I was annoyed. But once I showed it to a few other people, the reaction was always the same:
"Wait, why does this not already exist?"
That is usually a decent sign.
I am not pretending this is some giant category-defining startup. It is a very small product solving a very real annoyance for people who use AI tools all day.
If you are a developer, indie hacker, or anyone burning through Claude, OpenAI, or Cursor usage and wondering where the money went, that was the exact motivation.
TokenBar is live at tokenbar.site. It is $5 lifetime.
Mostly I built it because I got tired of surprise bills and I wanted the cost to be visible at the moment I was creating it.
That is still my favorite kind of product idea.
Small pain. Constant pain. Easy to explain.
If you have ever had an "oh no" moment with an AI bill, I would genuinely love to hear what triggered it.
Top comments (0)