Andrew Ng on the rise of AI agents and agentic reasoning

Why it matters

Andrew Ng — Coursera founder, former chief scientist at Baidu and Google Brain — is AI adoption's most patient evangelist. This keynote proposes four agentic design patterns — reflection, tool use, planning, multi-agent collaboration — often lifting performance more than a bigger model. In 27 minutes, why enterprises should start with agentic processes, not model parameters.

Key takeaways

• Stop debating 'what counts as an agent': design along an autonomy spectrum — fixed flows to highly autonomous.
• Four design patterns: reflection (self-critique and revise), tool use, planning (step decomposition), multi-agent collaboration.
• Agentic workflows let smaller models beat bigger ones used naively: process design is the new performance lever.
• Non-technical teams benefit too: break repetitive knowledge work into reflective, iterable steps.
• Adoption advice: pilot on well-defined, fault-tolerant scenarios first; expand autonomy only with data.