Conversations with docs is one possible application. I foresee conversations with papers, articles, books, the transcript of a YT lecture, or even with Google (

Conversations with docs is one possible application. I foresee conversations with papers, articles, books, the transcript of a YT lecture, or even with Google (oh no!). The other major LLM improvement I expect is multi-round code execution. Let the model iterate on bug fixing, like us. We need the extra steps, the LLM does too.