Since no new fundamental AI breakthrough has happened since transformers were invented, it stands to reason that all current applications of AI are just LLMs un

Since no new fundamental AI breakthrough has happened since transformers were invented, it stands to reason that all current applications of AI are just LLMs underneath. RAG is just an LLM plus regular procedural code acting on the LLM output. So are "reasoning" agents, because the reasoning is done via LLM predictions. "Agents" are just RAG applications that have access to your screen, keyboard, and mouse. It's LLMs all the way down.