Thanks for the recording, very interesting presentations and discussions !
On the topic of ChatGPT or LLM-based methods, I agree, it could be interesting to integrate a first level of questions answering based on initial data from ChatGPT + JMP Documentation to solve frequent and "easy" questions (or simply redirect to the corresponding pages/learning documentation), where users answers don't have a high impact (we often quote JMP documentation or redirect to it, so automatize this part would enable users to take part in better, richer discussions).
And in order to make sure this Q&A process is stable and in control, perhaps adding a feedback loop and ask if the proposed answer is helpful or not, to consolidate knowledge and flag possible errors or model's hallucinations.
