Andrew Ng co-founded Coursera and Google Brain, led AI at Baidu, and now runs DeepLearning.AI and AI Fund. His recent focus on “agentic workflows” — breaking tasks into iterative, tool-using loops rather than one-shot prompts — maps directly onto the kind of automation RunAgentRun covers. His weekly letter is a steady source of grounded, hype-free guidance for practitioners.
AN
Andrew Ng
One of AI's foremost educators and a leading voice on agentic workflows — turning frontier capability into practical patterns small teams can actually adopt.
What They're Saying Now
Pair CS fundamentals with agentic coding skills ↗
In The Batch, Ng urges developers who want to stay competitive to combine strong computer-science fundamentals with AI-assisted, agentic coding — not to treat them as alternatives.
The biggest predictor of agent success is disciplined evals ↗
Ng argues the teams that win at agents are the ones who run a disciplined process of evals and error analysis, rather than spending months tweaking prompts with little to show for it.
Build agentic workflows now ↗
His four design patterns — reflection, tool use, planning and multi-agent — remain the most practical recipe small teams have for building agents that actually work.
