Build and ship a web app with an AI coding agent
From idea to public URL with Claude Code: frame, plan then build, iterate with Git, publish from GitHub.
Course designed by Alex Dionisio, Senior Product Manager. Updated September 28, 2026.
- 24 steps
- ~2 hr 45 min
Level: Junior · Some basics are useful: introductory knowledge of the topic.
Go from a tool idea to a live web application, built with an AI coding agent (Claude Code first), without being a developer. You frame the need and your success criteria, have the agent plan and then build the app, improve it in small steps that are checked and versioned with Git, then publish it from GitHub to a public URL, knowing what is still missing before you hand it to real users.
What you will be able to do
- Frame a tool (problem, user, inputs, expected result) and write its acceptance criteria before generating a single line of code.
- Choose between a chat with artifacts, an app builder and a coding agent based on the need, the level of control required and the risk.
- Write a build request for a coding agent (context, scope, constraints, verification) and get a plan approved before any code is written.
- Set up the environment (terminal, Node.js, Git, Claude Code), run the application locally and explain what the main files of the generated project are for.
- Evolve and debug the application in small, checked, committed iterations, giving the agent the right context (symptom, steps, error message).
- Review the generated app before sharing it (behavior, edge cases, secrets, personal data, experience) and decide whether to ship, fix or stop.
- Publish the app from GitHub to a host with continuous deployment, get a public URL and evolve the live version without breaking it.
Prerequisites
- Able to install software and find a file on your computer
- Some experience with a chat assistant (Claude, ChatGPT…)
- Hardware: a macOS, Windows or Linux computer on which you can install software
- Accounts: a paid Claude plan for Claude Code (Pro or higher; the free plan lets you follow the "prototype" path with artifacts), a GitHub account and a Netlify account (free plans are enough)
- Recommended next course: Build an AI assistant for your product, to add an assistant to an app like this one.
Syllabus
What will I build in this course, and in what order?
- Course overview2 steps
Objective · By the end of this overview, you will know what you are going to deliver (a live web app, versioned on GitHub) and the four steps the course takes you through to get there.
What will I build, with which tool, and how will I know it worked?
Objective · By the end of this lesson, you will be able to explain what a coding agent does beyond a chat, and choose the right build approach for a need based on the control required, the risk and the cost.
Objective · By the end of this lesson, you will be able to describe your tool (problem, user, inputs, result, name) and write at least three checkable acceptance criteria before asking for a single line of code.
How do I get a first version running on my machine, and understand what was created?
- Set up your workstation2 steps
Objective · By the end of this lesson, you will be able to install and check your environment (terminal, Node.js, Git, Claude Code), create your project's versioned folder and start a Claude Code session in it.
Objective · By the end of this lesson, you will be able to write a five-block build request for a coding agent, get a plan approved in plan mode before any code is written, and obtain a first version running locally.
Objective · By the end of this lesson, you will be able to explain what the main files of a React and Vite project are for, say where your app's data lives, and deduce what it can and cannot do without a back end.
How do I evolve the tool without breaking it, and know whether it is ready to share?
Objective · By the end of this lesson, you will be able to evolve your app in short iterations (observe, describe, have it made, check, commit), review what the agent changed and go back to a previous version with Git.
Objective · By the end of this lesson, you will be able to write a complete debugging request (symptom, steps, expected result, error message), read the browser console, and review your app with a five-angle grid before deciding whether to share it.
How do I put the app online cleanly, update it safely and know what is missing for real users?
Objective · By the end of this lesson, you will be able to push your project to a GitHub repository without any secrets in it, connect it to Netlify with the right build settings, and get a public URL that updates on every push to the main branch.
Objective · By the end of this lesson, you will be able to ship a change through a branch, a pull request and a preview, connect a domain if needed, and list what separates your live app from an app that is ready for real customers.
Glossary terms
Public definitions of the key concepts in this course:
Guide on this topic
- Claude Code for Product Managers
What a PM can do with Claude Code, the method and the guardrails, with a starter CLAUDE.md.
Verified sources
The fast-moving facts of this course (tools, models, regulation) are checked against these sources, then reviewed according to how quickly they go stale.
- Claude Academy, Claude Code 101 · verified September 28, 2026
- Claude Code, Best practices · verified September 28, 2026
- Claude Code, Overview · verified September 28, 2026
- Claude Help Center, What are artifacts and how do I use them? · verified September 28, 2026
- Claude, Pricing · verified September 28, 2026
- Cursor · verified September 28, 2026
- GitHub Docs, About Git · verified September 28, 2026
- Lovable (company), Wikipedia · verified September 28, 2026
- Netlify Docs, Create deploys · verified September 28, 2026
- Netlify, Pricing · verified September 28, 2026
- Node.js, Download · verified September 28, 2026
- Node.js, Previous releases · verified September 28, 2026