“Early agent frameworks like Auto-GPT failed to gain traction because the underlying language models and the surrounding scaffolding were not sufficiently advanced at the time.”