The collapse of my last company was abrupt and turbulent. It left me with something I hadn't had in years: time. Time to unwind, and time to finally test the technology I’d been pushing our team to adopt—AI-assisted development. Many of the engineers I worked with had shunned it. “It’s good for generating automated tests, and that’s about it,” was the common refrain. I was a skeptic of their skepticism. After a few months of decompression, traveling to my favorite places, I finally dove in. What I discovered was that AI coding (aka. Vibe coding) had advanced far beyond even what I had imagined, proving the naysayers wrong.
The Myth of the Perfect Prompt
Everyone told me prompt engineering was an arcane science. To get the right results, they said, you needed to carefully craft context, instructions, and examples. Writing a quick email is one thing, but developing an entire application is another.
They were wrong—or at least, they were already behind.
Sure, if you ask an AI to generate a complete website from a single sentence, it won’t give you exactly what you want. It can't read your mind... yet. The process is iterative. It’s like pair-programming with a brilliant, hyper-literal junior engineer who needs clear guidance. Sometimes you get stuck in a loop where fixing one thing breaks another. But that doesn’t mean AI coding is flawed; it means your instructions weren’t clear enough. Provide the right instruction, and bam! A fix that once required a full day of work now takes half an hour of monitoring and approving an AI agent’s work.
"But Can You Really Trust AI-Generated Code?"
Some people ask how I can trust code written by an AI. My retort is simple: how can you trust any code your team wrote? Humans make mistakes. AI can "hallucinate" and provide wrong solutions. So do people.
The difference? Speed. I can now say with confidence that what used to take a team of five a full month to build, I can build alone in a single week.
An Extinction Event for Junior Developers?
A computer science degree from a good university used to be a golden ticket to a cushy job. But I personally know recent graduates, diploma in hand, who are struggling to find work. It makes perfect sense. AI-assisted coding has drastically reduced the need to hire and train junior engineers. A single seasoned developer using AI can now produce more, faster, than an entire team once could.
Frankly, I feel terrible that this is happening. Talented individuals who bet their livelihood on computer science are facing an unexpected and difficult reality. But this is a symptom of the world we live in—a world of rapid, transformative change. My hope is that this is a temporary, albeit painful, adjustment period. As the industry evolves, new roles that we can't yet imagine will emerge, and these bright minds will be the ones to fill them.
This evolution is no reason to vilify the technology.
The Danger of Standing Still
The shift to an AI-augmented life isn't just coming; it's accelerating exponentially. Staying on the sidelines isn't a neutral position—it's a decision to fall behind. In previous technological shifts, the learning curve was more forgiving. You could afford to wait and see. This time is different. The gap between the early adopters and the laggards is widening at a blistering pace.
Think of it this way: choosing not to engage with these tools today is like deciding to learn to swim during a tsunami. By the time you feel the water at your ankles, the wave is already towering over you. Those who are adapting now aren't just learning a new skill; they are positioning themselves to define the jobs of the future. Those who wait will find the landscape completely reshaped, with their old roles automated and the new ones requiring a fluency they never developed. Being dazed isn't an option; it's a path to obsolescence.
I don’t know what the future holds, but I know this: I have to stop being dazed and start adapting. We all have to learn the new way of doing things if we want to ride this wave instead of being swept away by it.