The basic architecture of an LLM-powered application: prompts, context, completions and tool calls
Open on Cached Info