Agentic Software Engineering: From Prompt to Production

Two days. Every developer builds, governs, and ships a working product.

Two days. Every developer builds, governs, and ships a working product.
Most developers already have access to an AI coding agent and most are only using it as faster autocomplete. That’s the smallest return available on a tool that can do far more. Agentic Delivery for Developers is a two-day, live virtual course where every developer moves beyond autocomplete and learns to build, govern, and ship real software with an AI coding agent at the center of the workflow.

Description

This course takes developers through the practices that make AI-written code trustworthy, in the order a change travels: briefing the agent, turning a story into a spec, writing tests first, working in small steps, running a quality gate, running the product live, agreeing what “done” means, and measuring delivery.

The two days are split accordingly: day one covers shaping a change and building it small, and day two covers proving it and shipping it.

Hands-on time grows through the two days, from about half of each session to nearly all of it.
Because AI coding tools change every week, this course is built around practices that outlast them. Developers leave knowing how to use today’s tools, and how to trust what any of them produce. Instructors demonstrate with Claude Code, and any licensed coding agent with desktop access works.

Who It's For

Working developers who build web applications and are comfortable with git and a terminal. No prior AI tooling experience is required, and cohorts that mix skeptics with early adopters tend to do the best work.

What You Leave With

• A deployed, working product that you can take with you, built from a starter backlog and still running after class ends.

• Project instruction files that hold the agent to your team’s conventions on every session.

• A pull request review gate you wire and test yourself, with a defect you plant on purpose.

• Tests written from acceptance criteria before any implementation exists, how a claim of “done” stays honest when nobody typed the code.

• A Definition of Done and a handover note, so someone else could run and trust your app.

Prerequisites

Every lab runs on your own laptop, so you’ll need a machine you can install software on. Plan about 45 minutes for setup, and finish it at least two days before class so licensing or IT approvals have time to clear.

Before class, please have the following installed and working:

• An IDE you’re comfortable working in that works well with AI coding tools, such as VS Code (current stable release).

• Node.js version 26 or higher, and npm version 12 or higher.

• An active, licensed AI coding agent, on the account you’ll use in class. Instructors use Claude Code.

• The desktop app for your agent, installed and signed in.

• Docker Desktop, installed and running.

• Git, with your user name and email configured.

• A GitHub account and the GitHub CLI (authenticated).

• A free Vercel account linked to GitHub, plus the Vercel CLI (authenticated).

• A free Resend account with one API key created.

• Zoom, with your camera, microphone, and screen sharing tested.

Using a Company Laptop? If your machine is managed by IT, check three things early: local install rights, outbound access to vendor sites, and whether a proxy or VPN inspects your traffic. These can take days to resolve. A personal laptop works fine if IT can't clear things in time.

**Registered participants get a detailed setup guide with step-by-step instructions and a verification checklist.