[RoboTeddy]

> but LLMs can’t do anything, really

Do you think https://github.com/Significant-Gravitas/Auto-GPT et al will become more performant as models improve?