Editor's Synopsis
Andrew Ng highlights a significant shift in how large language models (LLMs) are being developed. While LLMs have traditionally been optimized to answer users' questions — a paradigm popularized by ChatGPT — there is now a growing trend toward optimizing models to integrate into agentic workflows. Ng argues that this evolution will deliver a substantial boost to agentic AI performance.
This observation points to a broader industry movement where AI models are not just reactive question-answering tools but proactive agents capable of executing multi-step tasks, making decisions, and collaborating within larger automated systems. The shift from conversational AI to agentic AI represents a key inflection point in how these models are designed, trained, and deployed.