A new generation of prompts
this is going to change the current understanding of prompts and use the Agent as a smart script to be as predictable and precise
why ?
high token count but precise and expected output
less thinking
more generated content
in the experiment below :
130k+ token count on load
15s thinking
200s generating
cons: have to reset the session after making a checkpoint, which is about every response
google.dev wouldn’t let me share pictures because it’s my first day here😭, but i can share it elsewhere, I’ll even provide a video of how it works
but i need help on this project