A new package would process text inputs, like video titles or descriptions, to generate structured summaries using an LLM. It would take a user-provided text string (e.g., a headline or query) and ret
