Back to home
Learn
The failure modes of linear AI chat — and the workflows that fix them. Written for developers who build with LLMs every day.
What is context pollution?
Why long AI chats degrade — attention dilution, error compounding, and the structural fix.
8 min readContext branching: a developer's guide
Git-style branches for LLM conversations. The mental model, the mechanics, and four patterns that pay for themselves.
9 min readHow to run multiple LLMs in parallel
A fair model bake-off without juggling three subscriptions — method, evaluation criteria, and economics.
7 min read