Agentic Engineering

Introduction

Code agents can write large amounts of code in a short time. The hard part isn’t getting them started — it’s standing behind the result afterward. In this course, you’ll learn to delegate development work to AI agents in a way you can defend to architects, security, and audit.

When a colleague sends you code to review, you can ask about the choices behind it. With agent-generated code, you have to find the answers yourself — while there’s often far more code to review. This is where many teams struggle to keep quality up.

The course treats agentic engineering as an engineering discipline: what to delegate, what grounding the agent needs, how to verify the result, and how to document all of it. You’ll work hands-on, first on a shared case, then on your own codebase.

What you’ll leave with

• An overview of what a code agent actually does, and where it typically fails
• A template for structuring development tasks so an agent can solve them
• An agent setup you can put to use in your own codebase
• A checklist for reviewing AI-generated code
• Reusable prompts and workflows you can share with your team
• A concrete plan for how your team can start using code agents

Who it’s for

This course is for you if you’re a software developer, tech lead, or architect. You should be able to read and review code in the programming language you primarily work in. No prior experience with AI agents is required.

Work with your own stack

You’ll work with your own programming language, your own stack, and your preferred tool (e.g., Claude Code, Copilot, or Cursor). The course covers what cuts across tools: setup, context, prompting, harness, and agent loops.

On day 2, you’ll work in your own codebase. Check with your employer beforehand that you’re allowed to use an AI tool on it. If that’s not possible, we’ll provide a codebase for you.

Content and format

The course covers these topics: the code agent as a tool, research → plan → implement, spec-driven development, context engineering, reviewing AI-generated code, security and compliance, and going from individual to team.

The course is hands-on, with few slides and plenty of time at the keyboard. You’ll work alongside other participants and share what you’ve built along the way.

On day 1, you’ll solve new tasks in a shared codebase.

On day 2, you’ll work in your own codebase on concrete tasks from your backlog.

The course wraps up with you consolidating what you’ve learned and drafting a plan for taking the work forward.