How do you actually know what is burning your AI coding budget?

by

We built CodeBurn because we kept running into the same problem.

We run Claude Code, Codex, and Kimi across a few projects, and at some point the usage gets high enough that we want to know what caused it.

And that's where things get frustrating.

A typical day looks something like:

  1. Work normally across a few projects

  2. Switch between models and tools depending on the task

  3. Hit a usage limit, or see the invoice at the end of the month

  4. Try to work out what actually consumed it

  5. Realize the answer isn't really there

I want to know things like:

  • Which project is using the most?

  • Which model is responsible?

  • Was it one huge session, or hundreds of smaller ones?

  • Am I paying for long context, retries, or tool calls?

  • Am I using an expensive model for work that didn't really need it?

  • Which parts of my workflow are actually efficient?

The annoying part is that these are exactly the questions you want answered while you're working, not after you've already hit the limit or received the bill.

And once you're using several coding tools, it gets worse. Each one has its own usage view, terminology, and limits. Your workflow spans all of them, but the dashboards don't.

So I'm curious what other people actually do.

  • Just watch the usage percentage and hope?

  • Keep a mental rule like “cheap model for routine work, expensive model when it matters”?

  • Cap yourself somehow?

  • Start fresh sessions more often?

  • Actually measure usage and change how you work?

  • Or have you just accepted AI usage as part of the cost of the workflow?

We built CodeBurn to answer this for ourselves.

It reads the session data these coding tools already store locally and reconstructs the usage across projects, models, sessions, and tasks, so you can see where the usage actually went instead of just getting one big number.

But I'm genuinely curious about the problem itself.

How do you currently figure out what part of your work is burning your AI budget?

And what do you wish you could see that you can't today?

19 views

Add a comment

Replies

Be the first to comment