I use this actively when exploring new ideas to give me a quick, well rounded report and qualitatively is more amenable to my style vs simply prompting into Claude/ChatGPT for the same thing.
Thinking along these lines I’m also very interested in using LLMs in a similar fashion to simulate the propagation of memes / public discourse, in particular in response to arbitrary events. What the market considers “value” changes over time, and there is likely value in predicting how discourse evolves.