When your models spin up agents on their own by your instruction, it pays to know the balance of cost to ability for assignment of the tasks to the right models, even opting in cases for your own local less powerful but super cheap (free) task completions. Really enjoy some of these channels that are building multi-genetic dev systems with model agent chains and OS managed frameworks. Working on my own structure using some of their tactics. It’s a rabbit hole folks. Can’t just stand at the rim. Have to dive in. 🥸







