Usage of fluid compute

Hi all, I am using vercel for quite some time. Since fluid compute was intruduced it was automatically enabled for my app (not sure how). I am now getting closer to 4h of fluide compute in my hobby plan. I am interested how to enable it only for certain functions and when not enabled to use regular lambda function. What is very strange most of usage of fluid compute comes from middleware where Clerk middleware is used (50% usage). Everything else comes from small functions which I would like to run as regular lambda and for fluide use some longer jobs (analytics job fro example). Can you please provide more information how you defined if fluide is used for each endpoint/server action?

Thanks! :blush: