Try out the Codelab → https://g.dev/ai/adk-2-orchestration
Learn about Graph Engineering and how it provides the predictability and control needed for complex, multi-agent AI systems. In this conversational breakdown, we cover the differences between:
* Agent harnesses, loops, and graphs
* Walk through an automated pull request (PR) review pipeline using fan-out, join, and router patterns.
* How graph engineering differs from knowledge graphs, loop engineering, and agent swarms, and how to orchestrate shared state using Google’s Agent Development Kit (ADK).
Chapters:
00:00 – Intro: What is graph rngineering?
00:29 – Demystifying terms: Harness vs. loop vs. graph
01:29 – Real-world example: PR review with fan-out, join & router patterns
03:17 – Graph engineering vs. knowledge graphs
03:44 – Graph engineering vs. loop engineering
04:19 – Graph engineering vs. agent swarms
Speakers: Tilde Thurium, Annie Wang
Products Mentioned: Agent Development Kit, Graph Engineering
