after reading anthropic's latest post titled 'building effective agents' i thought it was time for me to finally, finally, finally get it over with and write out my thoughts on the subject. within crypto circles, there's been a lot of debate, and its been long enough that i need to make my voice heard. i spent most of my summer oscillating between crypto and AI + ML research so i feel as qualified as any other crypto bro to share my two cents.

anthropic's post (link here) was pretty good, though a bit boring as they mostly focused on the technicals of how to build these agents, the types of systems to accomodate agents, and how to properly use LLMs so you arent creating an agentic system just for the sake of saying you did.

this is just a small post and shouldnt be viewed as the entirety of my thoughts on the topic - i havent even gotten the chance to write about AI yet, and most of that writing stays within my obsidian files and never sees the light of day. the purpose of this is to really look at where the "agent ecosystem" as ill refer to it is today and where its headed. ill also avoid focusing too much on crypto twitter's agent ecosystem, as for the most part its just a bunch of automated twitter reply guys with a crude finetuned LLM underneath.