letting the LLM decide what action to take

splitting a big task into several smaller subtasks with JSON for the LLM

outputting JSON (instead of text) as a response to a prompt

using the output from an LLM for another prompt (and so on)

a bare-bones ChatGPT (or chatbot wrapper) on your own computer

pulling up with older words that sound more "rooted"

keying in some words to understand them better