Context optimization layer for LLMs. 65-90% token savings with zero quality loss. Drop-i…