A practical course on the coding agents, loops, harnesses, and governance patterns that make a software factory possible.
Agentic software delivery operates across three nested levels. This course covers all three, from the highest-level lifecycle changes down to the engineering primitives that make agents reliable.
| Level | What it covers |
|---|---|
| Agentic SDLC | How software delivery changes when AI agents participate directly in planning, coding, testing, and review |
| Agentic engineering | How engineers work inside that model — designing loops, managing state, and setting scope boundaries |
| Harness engineering | How agents are made reliable enough to participate at all — constraints, verification, and continuity |
AI coding agents — tools like Claude Code, Codex, and Cursor — are increasingly capable of writing, reviewing, and refactoring code. But capability alone does not produce reliable delivery. Without deliberate engineering, agentic workflows introduce new failure modes: context loss between sessions, scope overreach, silent errors, and accumulated drift.
This course rethinks software delivery for a world where AI coding agents participate directly in planning, coding, testing, and review. It is not about replacing developers. It is about designing the systems, constraints, and feedback structures that make human-agent collaboration work in practice — at the feature level, the session level, and across the full delivery lifecycle.
By the end of this course you will be able to:
We are preparing a new set of advanced chapters for engineering leaders and practitioners working on reliable human-agent software delivery.
These may include:
Tell us which topics matter most in your context.