I think you’re very interested in why you think the chat interface to LLMs is not great, and why you think your model might be better. But you don’t need people

I think you’re very interested in why you think the chat interface to LLMs is not great, and why you think your model might be better. But you don’t need people to follow that chain of reasoning for them to see the place you ended up as being interesting. Show, don’t tell. And start with what the tool does, before you explain why it does it that way.