BLOG
Thoughts on AI memory, context compression, and building in the open.
LLM conversations lose context as they grow. Rohrpost solves this with capsule-based compression that cuts token usage by 70% while preserving what matters.