Context

Greg Eisenberg discusses the concept of graph engineering, explaining how it can enhance AI workflows by structuring tasks into manageable steps rather than relying on a single AI interaction.

Key points

• Graph engineering is a useful concept that helps design AI workflows as manageable processes rather than one giant chat. 0:38
• Graph engineering differs from prompt and context engineering by focusing on workflow design around AI. 1:29
• A practical example of graph engineering involves breaking down a research task into multiple angles and roles. 2:51
• Graph engineering uses basic vocabulary like jobs, arrows, and state to define workflows. 3:35
• Knowledge graphs and agent graphs are two types of graphs used in AI. 6:45
• Graph engineering is beneficial for complex tasks with multiple steps and sources. 9:59
• The process of graph engineering can start manually before automating it with tools. 13:37
• Graph engineering enhances quality by making workflows less dependent on perfect prompts. 20:06
• The goal of graph engineering is to create the smallest graph that improves work quality. 21:14
• Graph engineering creates a memory that enhances future workflows. 21:46

Quotes

More verdicts

• SKIPEveryone is saying SOFTWARE IS DEAD (LIVE Q&A)2.8/5 · 1h 30 min
• WATCHJack Dorsey's Buzz: The New Hermes Agent?4.3/5 · 38 min
• WATCHMarketing Agents Are Too Good Now4.6/5 · 37 min
• WATCHMost Valuable Skill of 2026: Managing AI Agents4.6/5 · 44 min