Why Graph Engineering will 10x your Claude/Codex

This video explains graph engineering — structuring AI work as connected jobs (nodes and arrows) instead of a single chat. Greg Eisenberg contrasts knowledge graphs (relationship reasoning) with agent graphs (workflow orchestration), gives practical examples (research, support, content, code), and recommends starting manually before automating. He emphasizes checks, human gates, and producing reusable memory from each run.

– Core idea: break tasks into planner, parallel researchers, skeptic, merger, and human approval.
– When to use: multi-step work with parallel paths, checks, or approvals.
– Practical path: draw the graph, run it manually, then add file checkpoints and automation tools.

Quotes:

Graph engineering is how you design the work around the AI, so the whole thing stops living inside one messy, giant AI chat.

The model that writes the answer also grades the answer — that’s like asking someone to write their own performance review.

Draw the graph before you automate the graph.