Romanoo

Before ChatGPT & Claude
After AI psychosis

He finally lets his walls down and pours his heart out in a vulnerable paragraph
Make it so she replies with nothing but a "k"

> Start using AI agents to outsource my thinking
Ended up thinking more. Thinking deeply about thinking how to think of the things I've never thought about thinking off. All in the hope to outsource my thinking but think I've played myself

Bought a Grok Heavy Sub
How the hell did team Elon build a better terminal coding agent than Claude Code and Codex
It even comes with an llm-wiki, which is hopefully a more polished version
Well... now... let me think on how to put it to the test

Really? Is that in your training data or were you in autocomplete mode?
ChatGPT: there by pattern-completing from your correction, not from verified knowledge.

What about letting agents use github it's own build-in kanban for project planning and team to be update
Yes, github has that

Expecting us to wear our headphones & talk to strong AI models while live sharing our screen for AI to max assist

Supermemory, Manual kb, Holographic, Mem0, Retaindb, Byterover, Honcho
OpenViking, Hindsight
Which memory for agents?
Handle GCP infra, bitcoin core, L1/L2 chain dev, matching engines
Been using Byterover (curated MDs) + llm-wiki. Which one do you advise and have you used