Contextual AI has launched the world's first instruction-following, state-of-the-art (SOTA) reranker designed to enhance AI ranking methods. This reranker allows users to provide explicit instructions to control ranking strategies, addressing enterprise needs such as deduplication and chronological sorting. The reranker has achieved SOTA performance on benchmarks like BEIR and is integrated into Contextual AI's platform for building specialized Retrieval-Augmented Generation (RAG) agents. It introduces prompt engineering to information retrieval, enabling more tailored and efficient AI applications. The reranker is available via API and is part of a broader trend in AI development, including comparisons with Graph RAG and integration with frameworks like LangChain, which are increasingly used to improve retrieval accuracy and reasoning in enterprise AI systems.