Comment by r00fus

I'm somewhere between "what could possibly go wrong?" and "LLMs are just glorified autocomplete and can't reason".

I wonder if agents would be empowered to have access to any meaningful systems.