Introduction

Headroom is the context optimization layer for LLM applications. Compress tool outputs, DB results, file reads, and RAG results before they reach the model. Same answers, fraction of the tokens.