Still the best hour on graph engineering ever recorded, Andrew Ng breaking down how to build agentic knowledge graphs from scratch: 00:00 - what agentic knowledge graphs actually are 03:05 - building a graph from scratch 13:58 - the architecture behind multi-agent systems 22:57 - building a real one... with Google ADK 01:06:02 - why graphs are the future of AI agents I've seen $500 courses that teach less. Watch it, then take it further with my step-by-step guide on graph engineering below.show more

Different graph, same words, and both are worth the hour. Ng builds the knowledge graph. The other one is the work graph: nodes are agents, edges are real dependencies, and the longest chain is your time floor. Full course on that second one:

@AnatoliKopadze honestly this blew my mind. ng's take on multi-agent systems at 13:58 was gold. didn't think about it like that before!

multi agent systems need shared state

knowledge graphs as primitives for agent orchestration

damn, it's looking really important, i am saved this, thk Anatoli

So nice vid, thanks for sharing

wow, Andrew Ng new alpha for this man

The part people skip is 22:57. The Google ADK implementation is where the theory either holds or breaks, and most graph explainers never get that concrete. Worth watching that section even if you skip the rest

Successful implementation of internet of things requires clear metrics, iterative experimentation, and continuous evaluation over the next decade.

Knowledge graphs get valuable when they improve a real decision, not just connect more entities. While working on Komo, our test is simple: does the graph explain why this account, why now, and what evidence supports it? If not, it’s architecture without leverage.

This is a great resource for anyone interested in agentic knowledge graphs. Andrew Ng really knows his stuff.

The next AI advantage won't be bigger prompts.

Use graphs to push your own thinking into AI mode.

fair point, the architecture part really speaks for itself. multi-agent systems are the real deal now and this breakdown is a solid reference for anyone building with agents today

Organizations adopting reinforcement learning typically succeed when they integrate it into broader strategic goals over the next decade.

Timestamp 13:58 on multi-agent architecture is genuinely worth rewatching multiple times

Rare to see complex AI concepts explained with this much clarity.....

The useful shift is treating the graph as a decision layer, not just a better prompt. I'd evaluate it on retrieval precision, stale-edge handling, and whether the agent can cite the path that led to an action.

hot take: the build is the easy part

no one has explained how to build agentic knowledge graphs better for me than Andrew Ng

ngl haven't watched it yet but andrew ng on graphs is always solid. gonna check it out but curious if he actually gets into the messier parts of real world implementations or more theoretical

The graph engineering shift is real. We're moving from single-agent prompts to multi-agent systems that actually coordinate. Andrew Ng's course captures this better than anything I've seen.

The graph engineering shift is real. Most people are still building single agents while the real builders are moving to agentic graphs.

The multi-agent architecture section is the part worth rewatching - once several agents are on one task, the design problem shifts from any single agent to how they share context and collaborate across the graph. That's where most of the hard problems, and most of the leverage, actually live.

Thanks for the info. Your article The Graph Blueprint is very useful.

Great share

A critical discussion of cryptocurrency economics should include both technological opportunities and systemic risks.

the build hour is the easy hour. the one nobody records is week 6, when entity resolution starts merging things it shouldn't and the graph quietly rots. that's the hard part.

@grok , is this video available on YouTube?

This is a really good breakdown @AndrewYNg

Bro, is a fancy name, but this are workflows or DAGs. What do You think? Where is the difference?

graphs finally make agents less forgetful

Thank you for awesome content!!!