Who's Behind This
I'm Chris — a programmer based in Sweden. I work as a developer in public service, and I've been around computers for the greater part of my life.
I was born in the early seventies, got my first computer (a Commodore 64) in 1983. That machine was magic — you typed commands and it did things. That feeling never went away. I've gone through every era of personal computing since: BASIC on tape drives, the early internet, the web explosion, and now into whatever this AI era turns out to be. I build my own PCs — for gaming, but also for development work. Programming has been both my profession and my hobby for decades.
AI is a relatively new interest, but it's taken over a significant part of my time. When I started experimenting with AI coding tools, I realized they were genuinely useful — but the material about them was either breathless hype or dismissive skepticism. Neither helped me figure out how to actually use these tools in my daily work.
So I started writing down what I learned. That became this site.
Why This Site Exists
AI Programming Manual exists because I wanted the resource I couldn't find: structured, practical, honest guidance on building software with AI. Not marketing, not theory — just clear explanations of what works, what doesn't, and where the real trade-offs are.
The site covers two audiences:
- Developers — Programmers who want to integrate AI into their existing workflow. The 20-chapter manual, the project tutorials, and the tool guides are written for people who already know how to build software.
- Vibe coders — People who build by describing what they want to AI, without a traditional programming background. The vibe coding guides help people who are new to building software but have clear ideas about what they want to create.
Both groups are figuring out the same thing from different directions: how to work effectively with AI as a building tool.
How This Site Was Built
This entire site — the design, the guides, the tutorials, the articles — was built by me working with Claude over about six weeks of evening and weekend sessions. It started as a translation of a Swedish manual I'd written, then grew into the 40+ page resource it is now.
I made every editorial decision: what to cover, in what order, for which audience. Claude did the heavy lifting on drafting, HTML generation, and maintaining consistency across dozens of pages. The process involved hundreds of conversational exchanges, constant iteration, and a fair amount of restructuring when early decisions turned out to be wrong.
The full story — what worked, what went wrong, and what I'd do differently — is in How This Site Was Built: A Developer and AI, Start to Finish. It's the most honest account I can give of what AI-assisted development actually looks like on a real project.
What You'll Find
- The AI Programming Manual — 20 chapters from first prompts to expert-level workflows, structured in four tiers from foundations through mastery.
- Project tutorials — Build real software with AI, step by step. Every prompt, decision, and review shown.
- Tool and workflow guides — Editors, CLI tools, TypeScript/React prompting, legacy migration, REST API development.
- Vibe coding guides — Getting started, learning paths, common mistakes, and a hands-on walkthrough for non-programmers.
- Articles — Shorter reads on how AI works, where things are heading, and behind-the-scenes case studies.
Editorial Approach
- Practical over theoretical — Real prompts, working code, concrete workflows. If it can't be demonstrated, it's not ready to publish.
- Honest about limitations — AI makes mistakes in predictable ways. This site documents those failure modes. The field guide on AI errors exists because people need that information.
- No hype — No "10x developer" claims. No promises about replacing anyone. Just what works and how to use it.
- Testable — Every technique can be tried immediately. You should see results the same day you read a guide.
Get in Touch
Found an error? Have a topic suggestion? Want to say hello? Use the Contact page. Content corrections are especially welcome — accuracy matters more than pride.