The Longest Year: Vibe Coding and the 4-Hour Turning Point

I was out on my bike the other day when a post flashed across my feed: Vibe Coding is officially one year old.

My immediate reaction was visceral. I was floored. I almost couldn't believe it. Only a year? In my head, I've lived an entire career's worth of product cycles in those twelve months. When I told my wife later that evening, she had the opposite reaction: "A year? I thought it was way newer than that."

That dissonance fascinated me. It feels like we're split into two buckets now. There are those who "drank the Kool-Aid"—the ones who went deep, leveraged the tools, and fundamentally changed how they interact with the world. Then there are the skeptics or late adopters who may see it as a passing trend or a minor efficiency gain. Both perspectives are valid, but they lead to two completely different versions of reality as I learned.

For me, the reality changed in a Toronto hotel room in May of last year.

The Hotel Room Moment

My wife was at a work off-site, and I found myself with a few hours to kill and a conversation with a psychologist stuck in my head. We had been talking about the privacy hurdles of using AI in clinical practice—specifically the strict privacy laws in Canada and the fear of leaking personally identifiable information (PII).

I sat down with my laptop and started a conversation with ChatGPT. I didn't just ask it to code; I had a discussion about the problem. How could we intercept a prompt? How could we redact it in real-time?

I'd heard the term "Vibe Coding" floating around. I decided to see if the hype was real. I downloaded Cursor, and in a roughly four-hour span, I went from a conceptual "what if" to a working prototype of what is now PII Shield.

I didn't sleep well that night. My mind was racing. If I could do that in four hours, the cost of testing ideas had just dropped by 99%. The barrier to entry wasn't "Can I afford a dev team?" anymore; it was "Can I clearly define the problem?"

A Year of Shipping

Since that night, Vervian hasn't just been a consultancy; it's been a laboratory. When I look back at the volume of work produced in this "Vibe Coding" era, it's staggering:

• PII Shield: Now in beta, moving from that hotel-room prototype to a market-ready tool for data residency and privacy.
• Spartacus: An analytics tool for sports teams to measure the "invisible" metrics—team chemistry and social dynamics.
• Receipt²: A solution for my own tax-season anxiety. It uses AI-driven OCR to bucket expenses and track year-over-year tallies from a simple photo.
• The Product Product: An AI-guided workspace that turns my product management workflow into accurate project estimations.
• Choice UI: An AI-assisted design system that sweeps codebases for "orphaned" components and forces them back into alignment—keeping designers in control of the PRs.
• LLM Trainer: Our internal engine for creating high-quality datasets to train models like Llama.

The "Working Code" Philosophy

I know what the purists will say. "It's not enterprise-grade." "It's vibe-heavy."

Here's my take: Working code wins every time. I'm not building fragile scripts; I'm building solutions on top of a solid foundation with trusted tech partners who handle the heavy lifting. I'm a product person at heart, and for the first time in history, the distance between a product vision and a functional tool is essentially zero.

We have active users. We have products hitting the market. We have a workflow that would have required a 10-person agency three years ago.

If that's what "Vibe Coding" looks like after one year, I'm both excited and terrified to see where we are in 2027.