The best vibe coders are already coders
Here's a much needed perspective to get the best out of your AI systems.
Dear Builder,
It’s tempting to say the dust around AI hype is settling. But I won’t fool myself to believe that’s true.
If anything, the noise is still pretty loud. New tools, new launches, new promises every week. The narrative hasn’t slowed down. But underneath all of that noise, something more grounded is becoming true and I really hope you’ve been paying attention to it.
The people getting the most out of AI are not beginners. They are professionals.
The best vibe coders are developers.
The best vibe designers are already designers.
The best operators using AI are people who understand operations already.
That early idea that you could skip foundational knowledge and still produce high-quality work with AI is absolutely not true so don’t think of replacing people yet.
And honestly despite the consistent layoffs, it was always a bit of a marketing illusion and it still is.
But we already know this, so what’s the point of this article.
If you would like to stay up to date on how you can leverage AI in building products and companies of the future, this is the place to be.
The $20 Shortcut That Was Never Real
At least not in the way most people thought it would play out.
At the start of the AI wave, the pitch was simple:
Why pay a developer $2,000 when you can pay $20 for an AI tool?
Why hire a designer when AI can generate designs?
Why build a team when an automation agent can replace them?
It was a compelling story. And it worked, because it reduced cost on paper. But what we’re seeing now is different.
AI didn’t replace the need for skill.
It amplified the people who already had it.
In simple terms, for large corporations, it reduced the required man power but for small startup teams, it added the cost of using AI to the cost of having the human.
In fact in some cases, you have to hire an AI expert to help with the integrations, usage and maintenance of the systems you are created. Why?
How much stuff can you really keep up with when you already have a job to do?
AI Doesn’t Fix Broken Systems
Another shocking but obvious reality is the fact that what you are already working with matters more than your enthusiasm to integrate AI".
There’s something I’ve been seeing consistently while working with clients:
AI doesn’t create leverage from nothing. It multiplies what already exists.
If your operations are messy, AI scales the mess. If your marketing doesn’t convert, AI just helps you fail faster. If your design taste is poor, AI produces more bad design.
You cannot optimize something that is fundamentally broken.
And that’s where a lot of people get it wrong. They’ve been sold a possibility that they can never properly produce in their own business and in their own context and now they are mad about the difference between their expected outcome and their current reality.
The Hidden Layer
AI is great at giving you outputs. But it rarely teaches you the invisible layers that make those outputs actually work.
In building web apps and automations, I have realized that things like:
Error handling
API security
Version control
Deployment workflows
Database architecture
are sometimes conveniently overlooked by these tools. So many time my clients would with great excitement build something with AI and believe they are unto something. But they have conveniently skipped multiple layers of technical context needed to keep the product working, viable or functional.
You can ask an AI to build a dashboard, and it will.
But when something breaks, or when you need to scale it, or when you need to secure it, that’s where the gap shows.
It makes it look good, and they it tells you to upload the HTML file on netlify or vercel. It’s until you try to do that you realize:
you don’t have a PRD,
you don’t actually have QA plans
you don’t have plans for version control
… and so on.
That gaps are experience that a professional would not miss.
The Real Advantage Is Not the Tool
The advantage is perspective.
I’ve seen this firsthand working with operators who are already strong at what they do. The difference in outcomes is not the tool they use, it’s how they think.
How they structure workflows
How they approach client acquisition
How they connect systems together
AI just becomes an extension of that thinking.
Not a replacement for it.
So What Should You Do?
If you’re trying to build with AI, you have two real options:
1. Work with someone who already understands the domain
This is the fastest path. It saves time, reduces risk, and leads to better outcomes.
2. Become that person yourself
But that means doing the hard part first:
learning the fundamentals before leaning on AI.
There’s no real shortcut here. This has always been the truth.
“AI will not replace professionals”
“Professionals who use AI will replace everyone else”
And the sooner you align with that, the faster you start to benefit from what AI actually offers.
I hope this provided a much needed reality check for you.
Cheers.
Sidenote: I finally started using Claude Cowork, and Claude Connectors this week and they are amazing.
I also has a brief moment where I had to build something while my claude subscription was inactive. lol. It was hilarious but I survived.
Board Update
No update today. Might have a board for workflows created soon.
Stay Connected
🐦 Follow me on Twitter/X for daily AI + design experiments