Top 3 Vibe Coding Tricks

What are your top tricks or techniques when building apps and other digital experiences with AI?

The foundation of any successful collaboration, whether human or digital, is mutual understanding and shared context. When onboarding a new developer, you provide essential information about the project's architecture, technology stack, coding conventions, and goals... we can call it "vibe coding" all we want, but to be effective we need more than just vibes.

The AI can handle the syntax of programming but we are still creative directors and problem solvers.

To work together effectively you need structure.

Tip One: Instructions File

Your AI assistant requires similar guidance to operate effectively within the unique ecosystem of your project. Neglecting this step forces the AI to make assumptions, often leading to suggestions that are misaligned with your standards, incompatible with existing code, or simply inefficient within your specific context. Establishing clear ground rules from the outset is paramount. A highly effective method for achieving this is through a dedicated project instruction file, often named instructions[dot]md or .ai-instructions, placed prominently in the project's root directory. This file serves as a persistent charter, a set of guiding principles that the AI can reference to ensure its contributions are relevant and appropriate.

Think of this instructions file as the project's constitution for your AI. Using simple Markdown, you can articulate the essential "house rules" that govern development.

Tip Two: Focus on Why Not Just How

Don't just ask the AI how to implement feature X, or to make the change for you. Instead, ask it why it chose a particular approach, what the trade-offs are compared to other methods, or to explain the underlying concepts. Successful long-term software development isn't just about getting code that works now, but understanding why it works and how maintainable it is. Using the AI as a Socratic partner deepens your understanding, making you a better developer, better equipped to debug, refactor, and extend the system later, even if the AI isn't available.

Now you are constantly learning as a dev... not a vibe coder anymore. You are becoming a software engineer.

Tip Three: Treat the AI as an Imperfect, Unreliable Intern (Not a Senior Dev)

Many users default to treating advanced AI like a highly skilled expert. A more productive (and safer) mindset is to treat it like a very fast, often knowledgeable, but fundamentally unreliable intern who lacks real-world context, makes things up confidently (hallucinates), and needs constant supervision. Give it small, well-defined tasks. Review its output more rigorously than you would a human colleague's code. This mental model prevents over-reliance, encourages verification, and helps you catch subtle errors or architectural flaws introduced by the AI.

I had a basic math error in my AI-powered Life Copilot.. and it was ruining the RPG elements I had built.

Eventually I asked the AI to show me all the calcs and explain them step-by-step.

Now I know better than to assume AI is an oracle. It's a tool. Trust but verify!

TLDR: Don't just vibe code... vibe GROW