How the academy uses the tools
Codex and Claude Code are equivalent examples of the same build and review method. Students choose one compatible agent for implementation, troubleshooting, testing, and documentation; OpenClaw remains optional.
AI agents course
Instead of teaching agents as abstract architecture, this path teaches the operating layer: what to automate, what to avoid, how to test the agent output, and how to explain the service without hype.
Codex and Claude Code are equivalent examples of the same build and review method. Students choose one compatible agent for implementation, troubleshooting, testing, and documentation; OpenClaw remains optional.
The course avoids regulated, sensitive, and irreversible first projects. Students start with fake or anonymized data and a human-in-the-loop review point.
Student artifacts
A workflow pain finder for selecting a realistic first agent use case.
A lead follow-up, inbox triage, support FAQ, or report assistant demo shape.
A review boundary that explains where a human checks the agent output.
A proposal and handoff note that are honest about scope and limits.
Practical guides
These guides turn the category into one narrow offer, a fake-data demonstration, and prompts you can use with your coding agents.
Use an AI agent where judgment helps, keep predictable rules deterministic, and package one useful workflow a buyer can review.
Read guideChoose a small workflow a real buyer can inspect before you spend days building anything.
Read guideStart with outputs people review, not autonomous actions that can create expensive mistakes.
Read guideSource-informed, not affiliated · non-affiliation note
These links are used for positioning and screenshot/source planning. The academy is independent educational training and is not officially affiliated with OpenAI, Anthropic, or OpenClaw.