💭 Bulk llm calls might be useful for fermi estimating / planning