I’ve been watching a specific genre of LinkedIn post with growing irritation.
You know the type. A C-suite executive who hasn’t touched a codebase in years (or ever) posts about their transformative vibe coding journey. They’re generating thousands of lines of code per day. They built an entire app over the weekend. They’ve never felt more productive.
The implication is clear: this AI stuff is so easy that even a busy executive can outcode their entire engineering team while simultaneously running a company.
I want to be careful here. The agentic development revolution is real. AI tools are genuinely changing how software gets built. I use Gemini CLI and Claude Code every day, and it has fundamentally altered my workflow. This isn’t a “get off my lawn” rant about how kids these days don’t know what real programming is.
But there’s a difference between tools that amplify expertise and tools that replace it. And right now, a lot of executives are confusing the two.
The Numbers Don’t Lie (But Executives Do)
Here’s what the research actually shows:
A randomized controlled trial studying experienced developers found that using AI tools made them 19% slower on real-world tasks. This despite the developers feeling 20% faster. The gap between perception and reality is striking. This data is older and I don’t trust those numbers to be true in 2026, but the sentiment is there. Seeing lines scroll by really fast doesn’t mean a lot outside of a movie set.
Meanwhile, Google’s internal study showed developers completed tasks about 21% faster with AI assistance. That’s meaningful. But it’s a far cry from the 10x (or more) claims flooding social media.
Simon Willison, who actually understands this stuff, puts it well: certain tasks can see 2-5x productivity gains for experienced programmers. But the key words are “certain tasks” and “experienced programmers.”
Quick prototypes? Absolutely faster. Knocking out a throwaway script? Maybe five minutes instead of an hour. But production code that needs to be maintained, secured, and scaled? That still requires someone who understands what they’re building.
The Hangover Is Already Here
The vibe coding delusion has already created its first wave of casualties. Roughly 10,000 startups tried to build production apps with AI assistants. More than 8,000 now need rebuilds or rescue engineering, with budgets ranging from $50K to $500K each.
The total cleanup cost? Estimates range from $400 million to $4 billion.
AI-generated code includes 2.4 times more abstraction layers than human developers would implement. That’s not efficiency. That’s technical debt accumulating at a rate that makes the code unmaintainable.
Now, to be fair, there’s a real argument that maintainability standards need to shift when AI can regenerate code faster than humans can read it. Maybe some code doesn’t need to be maintained in the traditional sense. There’s a balance to find here. But right now? We’re way off it. Most of these startups aren’t making a conscious tradeoff. They just don’t know what they don’t know.
And security? 45% of AI-generated code contains an OWASP Top 10 vulnerability. Missing access controls. Hardcoded secrets. Unsanitized input. The basics that any experienced developer knows to check for.
This is what happens when people who don’t understand systems try to build them with tools that generate plausible-looking output.
The Real Insult
Here’s what bothers me most about the executive vibe-coding flex.
When a CEO posts about generating 10,000 lines of code, they’re implicitly saying that what their engineering team does is just typing. That the hard part of software development is the mechanical act of producing characters in a file.
It’s not.
The hard part is understanding the problem. Navigating constraints. Making tradeoffs. Anticipating edge cases. Building something that doesn’t fall apart when real users interact with it in ways you didn’t expect.
As Willison puts it: “Our job is not to type code into a computer. Our job is to deliver systems that solve problems.”
An executive who claims to be shipping code while running a company is either:
- Not actually running the company
- Not shipping production code
- Creating future technical debt for their team to clean up
Pick one.
What We Should Actually Be Celebrating
The real opportunity with AI-assisted development isn’t making non-technical people into developers. It’s helping technical people deliver value faster.
Faster iteration cycles. Quicker prototypes. Less time on boilerplate. More time on the hard problems that actually require human judgment, then faster execution on the plans once they’re developed.
MVPs that used to take 9-12 months now launch in 8-12 weeks. That’s transformative. But it’s transformative because experienced teams can move faster, not because we’ve eliminated the need for experience.
The goal was always faster value. AI tools help us get there. But they help us get there by amplifying expertise, not replacing it.
A Modest Proposal
If you’re an executive tempted to post about your vibe coding journey, consider this:
Your engineers see those posts. They know what production code looks like. They know the difference between a weekend hackathon project and software that handles real users, real data, and real consequences.
When you claim to be outproducing them with tools they also have access to, you’re not inspiring them. You’re revealing that you don’t understand what they do.
Instead, maybe post about how your team is using AI tools to ship value faster. Celebrate their wins. Acknowledge the expertise that makes the tools useful.
The clicks will come on their own when you’re actually solving problems for users.
That’s always been the point.
Hero image created with AI. The irony is not lost on me.