Discussions

We want to use live streamsing. Ideally we would have three different personas. Each would respond differently / perform a specific task. I think this means each needs a specific system prompt / prompt for the LLM model? Is this correct approach?