Beyond the Prompt: Vibe coding that actually ships
Everyone is vibe coding now: describe what you want, let the AI write it, hit run, repeat. It feels like magic until the project grows. Then come the bugs nobody understands, a codebase that fights back, security holes, and an AI stuck going in circles. This workshop is about doing it differently. We'll look at how experienced developers use AI tools like Claude Code, Cursor and Copilot to move fast without giving up understanding or quality. You'll learn how to: • Plan before you prompt: write clear specs, break work into small steps, and give the AI the right context • Stay the architect: make the structural decisions yourself and let the AI handle execution • Review, don't rubber-stamp: read, test and question generated code before it lands • Avoid the classic traps: spaghetti code, hidden security flaws, dependency bloat and endless fix-loops • Build a repeatable workflow: project rules files, version control habits, and knowing when to code it yourself Through live demos and hands-on exercises, you'll leave with a practical method for building real projects with AI that are fast to ship, easy to maintain, and that you actually understand. For: developers and students who already use AI tools (or want to) and want better results than "it works, I don't know why.
Everyone is vibe coding now: describe what you want, let the AI write it, hit run, repeat. It feels like magic until the project grows. Then come the bugs nobody understands, a codebase that fights back, security holes, and an AI stuck going in circles. This workshop is about doing it differently. We'll look at how experienced developers use AI tools like Claude Code, Cursor and Copilot to move fast without giving up understanding or quality. You'll learn how to: • Plan before you prompt: write clear specs, break work into small steps, and give the AI the right context • Stay the architect: make the structural decisions yourself and let the AI handle execution • Review, don't rubber-stamp: read, test and question generated code before it lands • Avoid the classic traps: spaghetti code, hidden security flaws, dependency bloat and endless fix-loops • Build a repeatable workflow: project rules files, version control habits, and knowing when to code it yourself Through live demos and hands-on exercises, you'll leave with a practical method for building real projects with AI that are fast to ship, easy to maintain, and that you actually understand. For: developers and students who already use AI tools (or want to) and want better results than "it works, I don't know why.