Agentic RAG AI — more marketing hype than tech advance

We agree with this view. For Agent systems to be viable, LLM accuracy and reliability has to approach 99.9% to avoid compounding error when executing multiple sequential tasks.

We are currently far away from that point and there’s little indication of that changing anytime soon