Google just released a free 1-hour lecture from Chief Scientist Jeff Dean on full AI Engineering. How to go from one LLM to coordinating 100 AI agents: 01:45 - Build an LLM from scratch 17:22 - Learn how to actually use AI models 30:03 - Master prompt engineering 52:35... - Coordinate 100 agents as one human Most people are still learning how to prompt one model. Jeff Dean is showing the full stack: LLMs → Prompts → Agents → Agent Teams → Graphs Prompting is the old workflow. Coordinating agent systems is the new one. This is 27 years of Google AI compressed into one hour. Bookmark and watch it today Then read the full Graph Engineering guide below ↓show more

had a specific idea I wanted to test after watching this - tried it right away and it actually worked

love when it works right away

I've been searching for LLM and guide how to coordinate a lot AI agents more convenient and I finally faced this video

glad you found it

love your posts, dear Lunar

really appreciate that

insane

actually insane

Bookmarked. Jeff Dean explaining this in an hour beats three months of blog posts.

way better than endless blogs

saved this video in my list

good one to save

@BrianRoemmele

Very useful lecture save this

definitely worth saving

Building an LLM is solved. The real challenge is the last bullet: coordinating 100 agents. State management across distributed agent loops is the unsolved piece — most frameworks treat agents as independent, but production systems need shared context and failure isolation.

这讲座标题起得挺鸡贼：从零构建LLM到协调100个代理，听着像给普通人的完整指南，实际是Jeff Dean在展示谷歌内部的全栈肌肉。大多数人还在纠结提示词怎么写，人家已经琢磨怎么让一百个AI代理像人类组织一样协作——这差距不是努力能填平的，是基础设施和算力堆出来的代差。不过我倒觉得他真正想说的不是技术，而是提醒行业：当所有人都在卷单模型能力时，真正的护城河早就变成了系统工程能力。你还在学怎么跟一个模型对话，他已经开���设计AI的职场关…

Prompt - agent - team- graph is just distributed systems relabeled. Dean's Pregel paper from 2010 already had the graph model. LangGraph copied it. He is teaching his own paper back.

"100 agents as one person" and "27 years compressed into an hour" is the same inflation pattern every AI thread uses now. also that reply chain underneath is a textbook engagement bot — same account replying "glad you found it" to five different people in a row 🤖

the whole engineering meta changed thanks to AI

Love that Jeff's focused on automating discovery. That's foundational AI work. Curious what Discovery Loop will tackle first.

100 AI agents need 1000x the coordination overhead of one. Jeff Dean didn't teach you AI. He taught you how to manage chaos

Prompting is optimization for one model at a time Coordination solves for when the one model doesn't exist yet

Est ce qu'il y a un lien youtube pour cette video?

@tlo9966

harness vs graph for production-what's your pick?