Semantically create chunks from large texts. Useful for workflows involving large language models (LLMs).