Summary
Nate Herk delivers a comprehensive one-hour course on OpenAI’s Codex, framing it as a “super app” that extends well beyond the ChatGPT web interface by enabling persistent projects, local file access, browser and mouse automation, reusable skill-building, and scheduled background automation. The video is structured for both complete newcomers and users already familiar with Claude Code, with explicit comparisons drawn throughout between the two platforms.
Key topics include Codex’s project and workspace organization, the distinction between conversational chat and full agent execution mode, model selection across GPT-5.5 and 5.4 with adjustable intelligence tiers (low, medium, high, extra high), and the /slash personality toggle for switching between friendly and pragmatic response styles. Herk walks through a complete end-to-end project: pulling YouTube comment data via API, analyzing sentiment and themes in an Excel sheet, and configuring the automation to run on a recurring schedule — illustrating how Codex can operate as a persistent worker rather than a one-shot assistant.
The course also covers the side chat feature for asking parallel questions without interrupting an active agent session — conceptually similar to Claude Code’s /slash command — and addresses file organization strategies as projects grow in complexity. Herk is candid about each platform’s relative strengths: he finds Claude better for exploratory brainstorming and planning, while Codex excels at executing long structured plans, following multi-step instructions, and troubleshooting issues that Claude sometimes struggles to resolve.
📺 Source: Nate Herk | AI Automation · Published May 06, 2026
🏷️ Format: Course Lesson
