How do the best developers amplify their abilities with AI? It's not magic. It's process.
Vibe coding: point your AI buddy at a problem and let it churn out thousands of lines of code.
That's a recipe for disaster-like using autocomplete to write your entire next email.
Instead, treat AI assistants as tools. You plan, take small steps, stay in control, get feedback, and learn. Done well, you'll do in hours what once took days. From whipping up a proof of concept to breathing new life into a decade-old codebase, Uberto shows how professionals harness AI to deliver consistent, accurate results.
You'll learn the small, controlled steps to take, the instructions that set guardrails, the prompts that work, and the workflows that maximize value while minimizing costs-for individuals and for teams.
Don't just read about it-watch it. Companion videos show these practices in action as Uberto builds both simple and complex applications, handling real-world problems as they arise. See how a professional configures and guides AI tools to get the job done.
Build a prototype in a weekend or a full product in a month or two. Untangle legacy systems, improve tests and documentation, and tackle the messy, real-world problems that slow teams down. Let AI handle the simple and mechanical parts of coding so you can focus on the creative, critical bits, potentially learning new languages and frameworks along the way. Whether you're debugging, starting fresh, or juggling a dozen tickets, you'll learn how to collaborate with AI in ways that feel natural, powerful, safe, and genuinely productive.
Table of Contents:
- Foreword
- Preface
- Why This Book
- Who This Book Is For
- How This Book Is Structured
- Online Resources
- What’s Next?
- Going with the Vibe
- Vibe Coding a Game
- Improving Your Prompts
- Writing a More Complex Game with AI
- What Went Wrong
- Using Specifications to Drive Assistants
- What We Learned
- Building Securely and Predictably
- Keep the Edge
- One Prompt, One Commit
- An AI with Good Habits
- Building an Application Following Rules
- Results and Reflections
- What We Learned
- Learning by Coding
- The Inception
- Shaping the Idea
- Walking Skeleton
- Modeling the Domain
- Developing the Search UI
- Show PDFs
- What We Learned
- Working on a Large Codebase
- Managing Technical Debt
- Understanding Legacy Code
- Enhancing Big Code Projects with AI
- Is It All Worth It?
- Cheat Sheet for the AI Powered Developer
- What We Learned
- Collaborating Inside an AI-Powered Team
- Adopting AI
- Team Organization
- Security Risks
- What We Learned
- Demystifying Large Language Models
- Attention Is All You Need
- What Are Large Language Models?
- Vector Databases
- Considerations
About the Author :
Uberto Barbini is a passionate, pragmatic programmer and longtime advocate of TDD and Extreme Programming. He's worked with many languages, but favors Kotlin for its clarity and Python for its simplicity. Today, he helps teams integrate AI assistants into their workflows. He's an international speaker and the author of several books and open source libraries.
Review :
Code agents are everywhere, and they will not go away. Developers interested in trying them out should follow Uberto on his fascinating journey. This way, they will learn to use them not only for prototypes but for production-quality software.
— Maik Schmidt, Software Developer and Author
Uberto has done a fantastic job pulling together this guide for developers and engineering managers alike. Many of us are still scratching the surface of what AI assistance can and cannot do, and this is great first guide to bootstrap that process. Very much recommended reading for all within the development community.
— Jayesh Shavdia, Managing Director, Head of Markets Core Risk, Barclays
The tools we're using to create software is changing, but the principles of making systems easy to maintain are largely the same. This book provides balanced and pragmatic advice on how to use AI assistants to code faster without compromising quality.
— >Andrea Goulet, Co-Founder, Legacy Code Rocks
The perfect "step 1" for anyone looking to understand how AI is transforming software development—enjoyable, accessible, and refreshingly honest about what works and what doesn't.
— David Denton, Co-Creator, http4k
