Between March and August, our token usage at Infracost grew 4.9x while our AI bill grew 2.4x.

Model routing and compression barely helped. The bigger savings came from an internal agent that reviews every model call in our code and opens pull requests to fix waste. It has cut our product AI spend 45% so far, and in one case it found a model reading the same file 74 times per run. That workload now costs close to nothing.

Alistair Scott, our co-founder and CTO, is walking through how it works in a 30-minute live session, with time for questions at the end.

If you can't make it live, register anyway and we'll send you the recording.

The Infracost team

P.S. Bring your weirdest AI usage questions. Alistair will answer them live.