The long-chat point got me. You think you’re carrying useful context forward. Meanwhile the AI’s rereading the whole family history every time you ask one more question.
That's why I keep very specific tasks in different chats, so the prompt and context is brief.
Exactly. Good habit. I have it configured so it suggests me to move to another chat to optimize token expense. If agreed, it creates a handoff that runs on a new task
That single prompt alone is gold, and can make the difference between getting real work done, and getting stopped right when you're making progress, Juan!
The reread point is the one that genuinely surprised me when I audited my own, that a long session quietly re-reads its whole history on every turn while you sit there none the wiser. The thermos line for caching made me laugh, Juan, because it explains in one image why the obvious fix did nothing for daily agents. Of all five, which gave you the biggest drop once you actually changed it?
Really useful piece, Juan. Any serious time saving or workflow optimization through AI feels extremely relevant right now, especially because the real cost is often hidden in small repeated inefficiencies.
I liked the point that the problem is not always the AI itself, but what we keep asking it to read before it can do the work.
The long-chat point got me. You think you’re carrying useful context forward. Meanwhile the AI’s rereading the whole family history every time you ask one more question.
That's why I keep very specific tasks in different chats, so the prompt and context is brief.
Exactly. Good habit. I have it configured so it suggests me to move to another chat to optimize token expense. If agreed, it creates a handoff that runs on a new task
That single prompt alone is gold, and can make the difference between getting real work done, and getting stopped right when you're making progress, Juan!
Thanks Mike. Glad you found it worthwhile
The reread point is the one that genuinely surprised me when I audited my own, that a long session quietly re-reads its whole history on every turn while you sit there none the wiser. The thermos line for caching made me laugh, Juan, because it explains in one image why the obvious fix did nothing for daily agents. Of all five, which gave you the biggest drop once you actually changed it?
The morning brief. The most recurring one that is the brains of the operation. Here is where most of the decisions are made.
Really useful piece, Juan. Any serious time saving or workflow optimization through AI feels extremely relevant right now, especially because the real cost is often hidden in small repeated inefficiencies.
I liked the point that the problem is not always the AI itself, but what we keep asking it to read before it can do the work.