@Voxyz_ai Ahh yes - my plan is to set the team up on Mac Mini, but no claw on Mac OS, here is local LLM and other tools that can be securely exposed as services to agents, all running in Apple containers (tiny Linux) where they can rule and destroy independently, with diff. level off trust
@morganlinton It’s a very cool setup - did you experience if it’s teams OR subagent setting or can Claude choose to use a subagent if it’s fitting for the task? And how does the team agent manage contex fill up? If the pick new tasks themselves, without clearing contex?
@bcherny It’s awesome, already feeling the difference 🙌🏻
But you should use Claude yourself for business sales support, still waiting for response more than a week in 🥴
@waitbutwhy Nothing is changed with disruptive tech - driven by selected few, hard to use and understand, deployment is slow at first. But as the S curve hits, it’s changes everything.
The truly mind blowing in our time is the speed of iterations 🤯
@waitbutwhy Definitely mind blowing! But not real time: “Queen Victoria's message of 98 words took 16 hours to send”
And as always with 1st gen: “the line speed was poor, and efforts to improve it caused the cable to fail after three weeks”
Took 8 years to get new working cable in place.
@TobieDk@KevinJPetersen@skattefar@fbhdk Og her 10 år efter svaret fra Skat så er det stadig kun samme klods af en XML fil man kan hente. 99 Gb zipped and counting 😖