How to use headroom to trim tokens right before sending them to an LLM

When building a RAG pipeline, do you ever just dump DB query results directly into your messages? When creating a log ...