My client CentralSquare Technologies is finding that when they upload a project that has subquestions nested underneath a parent question, the AI Draft is unable to use the context provided in the parent question to correctly answer the subquestions nested underneath it (for them, oftentimes these subquestions are only 1 word because the entire context of the question is provided in the parent question). They want the ai draft to be able to recognize the parent question context to be able to correctly answer the subquestions.