Summary
Nate Herk breaks down 18 core concepts of OpenAI’s Codex for people with no coding background, organized into four parts that build from the basics to more advanced features. The video is a hands-on walkthrough inside his own working setup.
It begins with foundations. Projects are described as simply a collection of folders and files that give the AI persistent context, so users don’t have to re-explain their business in every new chat. The video then explains AGENTS.md, a plain-language rules file that Codex reads before every request, including a routing map that points the agent to the right information. Herk shows how he uses this setup as an AI operating system for his business.
Later sections cover local work trees, which let you test risky changes in a copy of a project, and the .codex folder, where project-level and global settings, memories, sessions, and automations live. Herk is candid about which features he uses constantly and which matter less for knowledge work, giving beginners a clear picture of what to learn first.
📺 Source: Nate Herk | AI Automation · Published October 03, 2026
🏷️ Format: Tutorial Demo
