I Started Watching AI Spend Like Battery Life

I didn’t build TokenBar because I wanted another dashboard.

I built it because I kept doing the same annoying thing every day: opening an AI tool, getting into flow, and then realizing way too late that I had no idea what that session was costing me.

That sounds small until it isn’t.

At first, AI spend felt like a monthly problem. Something I’d deal with when the bill showed up. But that’s not how it works when you’re a solo dev using models all day. The cost happens in real time. One long prompt. One messy refactor. One “let me just ask it one more thing” rabbit hole. Then suddenly you’re not looking at productivity anymore — you’re looking at a surprise bill.

I remember the exact feeling that pushed me over the edge: I was deep in work, bouncing between prompts, and I had this tiny panic in the back of my head that I was burning money without seeing it. Not enough to stop. Just enough to make the whole workflow feel dumb.

That was the moment I stopped thinking about tracking AI usage as reporting and started thinking about it like a live signal.

Not a spreadsheet.
Not a dashboard tab I’d forget to open.
A menu bar app.

Something visible all the time, the same way you glance at battery, Wi‑Fi, or time. Because token usage is basically the new RAM usage: it’s invisible until it starts hurting you, and by then it’s already part of your day.

That’s what TokenBar does for me now. It sits there at tokenbar.site, quietly showing me what my AI work is doing in real time so I don’t get surprised later. No ceremony. No “analytics.” Just a number I can actually use.

I didn’t need more motivation to build with AI. I needed less surprise.

That was the product.