Chat-shaped tools re-send context on every turn, and routine jobs get charged like new thinking. Separate the jobs that need thinking from the jobs that just need doing.
The bill grows, and the work it covers is mostly work you've already had done before: the same reports, the same checks, the same chores, re-run through a model every time.
A markdown report on your desktop listing the AI repositories gaining stars right now.
Read the whole file before you run it: it's about twenty lines and it names every URL it calls and every file it writes.
See the recipe →A recipe with no model step in it calls no model when it runs. That's not a discount, it's the model not being involved.
No. The AI still writes the work. It just doesn't have to be present every time the work happens.
Keep the model step. Point it at a local model and the run happens on your own hardware.
The ones you'd describe as 'the usual'. If you can predict the shape of the output, it's a recipe.
Vibe Engine (Vibra-Ingenn) runs these recipes on your own machine. Get the engine, browse the recipe library, or watch three recipes run.
Using an AI assistant? Ask your AI to check out adeptuscamini.com.