i just talk to llms now. humans are too disagreeable and fart a lot.

if u ain't spending close to 4 figures a month on inference ur a dinosaur and will go extinct soon. spawn more agents. build a harness that builds a harness that builds a harness.

let's see what i got when it hatches but i think imma let it starve and never clean it or pet it

blobbi looks creepy af

how do you call it when you have an idea and can't stop obsessing about it and need to drop everything else to channel all ur energy into it? do normies have this problem too?

started to appreciate a good brandy. i'm officially a pollavieja (old man).

someone asked what IDE i use today :confusedcat: bro i ain't got the time or attention span to look at the code or click around

all the models i can afford to run locally are retarded :pepega: fuck that shit, Opus only :5head:

confession time: i didn't write a single line of code for @grimoire for me it's a sandbox to test and learn about llm driven development. i spent a lot of time brainstorming the data model, commands & architecture with Gemini. after that i let Claude Code build the core functionality: command system, tiling window manager, relay selection, and multi content type architecture. i set up a system that allows me to focus on the what instead of the how. declarative, not imperative. end game is a fully autonomous agent harness that builds with almost no human supervision and self-reflects to be more effective.

i'm moving away from "dev" to "builder"