Vibe Coding Is Already Dead
A year ago I was using AI to write code and fix bugs. Useful stuff — maybe 2-3x faster on feature work (last year it became 10-15x). The term “vibe coding” made sense. You describe what you want, the AI generates it, you review, you ship.
That’s not what’s happening anymore.
I build identity and biometric verification systems. My day-to-day involves Go backends, Solana programs, Terraform infrastructure, Kubernetes, and a Next.js marketing site. 1.5 years ago, ~30% of that was AI-assisted. Now it’s closer to 99%.
But the interesting part isn’t the percentage. It’s what shifted.
The AI isn’t just writing code for me. It’s running my engineering operations. It’s plugged into my monitoring, my deployment pipelines, my error tracking, my project management, my team’s Slack. When something breaks in production, it investigates the logs, traces the root cause, writes the fix, deploys it, and pings the team. That’s not code generation. That’s process automation.
There were many tools we used that didn’t really support these types of automation. Instead of working around it, I just made new ones and host everything behind a company-wide Tailscale VPN.
I went from “AI helps me code faster” to “AI simulates most of what an engineering team does.” Product review, planning, development, testing, deployment, incident response — all of it. The whole loop.
The word “vibe coding” implies you’re casually generating snippets. That framing is already too small. What’s actually happening is AI replacing entire engineering workflows, not just writing functions.
Three things I notice:
The technical founder is back. One person with these tools ships at a pace that used to require a 10-person team. The constraint isn’t headcount anymore — it’s how well you’ve instrumented your systems for AI to operate within.
Tool integration is everything. An AI that writes code is nice. An AI that’s connected to your monitoring, deployments, project management, and communication channels is a different category of thing. That’s the actual unlock — not better code generation, but deeper system access.
The next wave is process automation, not code generation. The teams that figure out how to point AI at their internal operations — not just their codebase — are the ones pulling ahead right now. That’s exactly what I’m spending most of my time on.
Nobody calls spreadsheets “vibe accounting.” When a tool becomes infrastructure, the cute name dies.