Hello, I’m trying to understand the credit system because I’m honestly quite concerned about how quickly our credits are being consumed.
I understand from the pricing documentation that a normal flow run costs 1 credit regardless of how many steps it contains, but AI steps are charged on top of that: 2 credits for Fast, 10 for Smart, and 20 for Frontier.
What I find quite confusing is how this translates into actual usage. We have a flow processing emails, and most of the emails do not even reach the AI step, yet we have already consumed around 10,852 credits out of 12,000 in the last 30 days. The project itself shows 7,620 AI Usage credits and 2,266 Run Usage credits.
This is particularly surprising because we are processing emails, not running an AI-heavy workflow on every single email, not even 10% of our inbox reaches the AI step.
Is there any way to see a detailed breakdown of exactly which AI steps are consuming the AI credits, how many times each AI step has run, and when those credits were consumed?
The current Usage page doesn’t seem to provide enough information to understand where the AI usage is coming from. Without that breakdown, it’s very difficult to diagnose unexpected credit consumption or estimate how many emails a given credit allowance will actually cover.
10,000+ credits being consumed while the majority of our emails don’t even reach the AI step seems extremely difficult to reconcile from the information currently exposed in the dashboard.
Is there a more detailed usage log/API/report that we can use to investigate this?