A few months ago I kept losing track of how much I was actually spending on AI.
Not the estimate in my head. The real number.
That was the problem.
I’d finish a day of shipping with Claude, OpenAI, and a bunch of little prompts, then get hit with a bill that felt larger than it should have been. The cost was invisible while I was working, which made it easy to ignore.
So I built TokenBar.
It lives in the macOS menu bar and shows token usage in real time. No dashboard. No digging through billing pages. Just a constant reminder of what the work is costing while you’re doing it.
That sounds small, but it changes behavior fast.
When cost is visible, you write better prompts. You stop wasting cycles on junk. You catch runaway usage earlier. You start treating AI like a tool with a price tag, not magic.
That’s the pattern I keep coming back to as a solo dev:
- If something hurts and the pain is frequent, make it visible
- If the fix is cognitive overhead, put it somewhere impossible to miss
- If a product can reduce decision friction, it will get used
TokenBar is not trying to replace your billing dashboard. It is trying to keep surprise bills from sneaking up on you in the first place.
If you’re building with AI and you care about spend, that’s the point.
TokenBar is here: https://tokenbar.site
Top comments (0)