Why we built the Playbook

There is a persistent gap in every product course: students leave with validated ideas, refined user personas, and polished pitch decks — but no working software to show. The jump from "I have a great concept" to "here is a prototype you can click" has traditionally required months of learning React, state management, build tools, and component libraries.

We built the Agentic Engineering Playbook because that gap is now unnecessary.

What changed

The last two years of AI coding agents have made it genuinely possible for a non-programmer to generate a working React prototype from a product document. The catch is that "possible" and "reliable" are not the same thing. Most attempts stall because:

1. The agent does not have enough context about what to build
2. The student does not know what to say when the agent asks clarifying questions
3. The generated code runs but looks like a placeholder wireframe

The Playbook solves all three problems. It teaches students how to write product documents that give the agent enough signal, how to answer its questions confidently, and how to iterate when the first output is not quite right.

Who it is for

The Playbook is for students in the Agentic Engineering course at our university — specifically students who are not developers and have never written production code. The five-step guide is designed to be followed in a single afternoon lab session.

But we also wrote it for AI agents. Every page ends with a machine-readable callout pointing to the next page, and the full guide is available at /llms-full.txt in a format that any agent can ingest in one request.

What comes next

The Playbook is version one. We plan to add case studies from real student projects, a troubleshooting FAQ, and guides for more advanced customization once the agent has generated the base prototype.

If you have feedback, open an issue on GitHub.