Built, Not Prompted

Stop reading Claude Code tutorials. Start building.
There is a moment when you are building with Claude Code where something looks right, compiles, runs, and is quietly wrong.

If you have been there, you already know why most content about this tool does not help you. The five-minute videos, the prompt-collection threads, the breathless productivity claims: they all skip the moment where plausible stops matching correct. They treat AI-assisted development as a magic trick. This book treats it as engineering.

Built, Not Prompted is a short, focused guide for developers who want real answers to the questions nobody asks out loud.

What this book covers
The book walks through building a complete event-driven microservices system in .NET 10, from an empty folder to five services running with asynchronous messaging, using Claude Code as the pair programmer throughout.
Along the way, it answers the questions that shallow content skips:

• If CLAUDE.md is always loaded, when is a Skill loaded? This book explains the actual mechanics of how Claude Code sends context to the language model, and why Skills scale where putting everything in CLAUDE.md does not.
• Why did the AI produce plausible output that was quietly wrong? The book documents a real verification moment and teaches the instinct that separates professional AI-assisted development from rubber-stamping AI output.
• When should I capture a pattern as a Skill? The book shows exactly when repetition becomes a signal, how to capture the pattern, and how to test that the Skill actually fires on the next task.
• How do I review AI-generated code without just rubber-stamping it? Verification is treated as the primary discipline of AI-native development and taught by example, chapter by chapter.

What makes this book different
Every prompt shown was actually typed. Every piece of code was actually generated by Claude Code. Every confusion documented was actually encountered. Every clarification was earned in real conversation, not retrofitted afterward.
The author brings four decades of experience across Assembly, C, C++, C#, database platforms from DBase to SQL Server, cloud architectures, and now AI-assisted development. The perspective is that of a senior engineer who has watched software reinvent itself multiple times.

What you will build
By the end of the book, you will have a complete, working system on your machine:

• Five .NET 10 microservices (Customer, Product, Order, Shipping, Notifications)
• Each with its own SQL Server database
• Communicating through RabbitMQ via MassTransit
• Orchestrated by Docker Compose
• A reusable Skill that makes adding future services trivial

The technology choices are specific. The patterns are universal. Whether your daily work is in Python, TypeScript, Java, or anything else, the rhythm of AI-native development stays the same: start with a clear blueprint, work incrementally, verify carefully, capture patterns, let experience compound.

Who this book is for

• Software engineers, technical leads, and architects who want to use Claude Code seriously
• Developers who are skeptical of AI hype but open to AI leverage
• Anyone who has read a Claude Code tutorial and thought "that was fine, but I still do not really understand what is actually happening under the hood"

If you build software for a living and want to understand the mechanics of AI-native development, not just the marketing around it, this book was written for you.

Built, Not Prompted: A Guide to Claude Code. The honest guide to building with AI.