AI Coding Tools Amplify What You Already Know
Alex MacCaw’s recent post about “vibe coding” for senior engineers is completely right. His core insight:
“If you know what you’re doing, have a deep understanding of the frameworks and libraries, and a clear idea of the way you like to do things, Vibe coding is like adding Nitroglycerin to your productivity.”
This hits because these tools amplify what you already know. They’re productivity multipliers for people who understand the domain.
Tool Choice Matters Less Than Commitment
I prefer Claude Code over Cursor, but honestly the tool matters way less than committing to one and learning how it works. You could spend forever evaluating new tools and never get real work done.
Pick one and stick with it long enough to get good.
What Actually Works
Two things from MacCaw’s post really clicked:
⚡ Quick iteration loops. “Write a little code, lint it, run the tests, and then iterate. AI doesn’t tend to do this by default. The models are like over-enthusiastic junior engineers.” You need to guide the process. Break things down, test incrementally, iterate based on what you see.
🎤 Audio prompts work. They’re messy and unstructured, but models parse them fine. They naturally lead to longer, more detailed prompts which get better results. I use audio for most of my Claude Code interactions now.
Six Prompting Principles That Apply Everywhere
MacCaw’s six prompting principles work beyond coding. These match what I teach in workshops:
🎯 Always start with a plan - Tell the AI what you’re trying to accomplish before asking it to execute
📝 Be specific about output - Define exactly what format, length, and style you want
💡 Give examples and context - Show the AI what good looks like with concrete examples
🚧 Use constraints liberally - Tell it what NOT to do, set boundaries and limitations
🔍 Keep scope tight - Break big tasks into smaller, focused requests
🗣️ When in doubt, ramble - Longer, detailed prompts consistently beat short ones
Whether you’re writing code, analyzing data, or creating presentations, these work. The AI needs context and constraints to deliver what you actually want.
Why This Works for Senior Engineers
You can quickly spot when code is approximately right. You understand how pieces fit together. You have opinions about what good looks like. When you can articulate what you want clearly, AI tools produce results that match your standards instead of generic implementations.
The productivity gains are real if you invest time to learn how these tools actually work. Senior engineers who figure this out will have an advantage - not because AI replaces expertise, but because it amplifies what experienced people can accomplish.
Read Alex’s full post - it’s worth your time.