A few weeks ago, I was standing in front of the team during our usual Wednesday meeting. I shared six points about the next four years. The first one went like this:
The classic web agency model is no longer sustainable.
It’s hard to admit that what we’ve excelled at for years no longer carries the same weight in the market as it used to.
When the camera starts to zoom out
We’ve been working together for years in a predictable routine. Projects come in, processes run smoothly, and everyone knows their role. For a long time, I didn’t question this either. We talked about our services using the same words as we did five years ago. But in the meantime, I realized that the world now understood something different by them.
There was an image that kept coming back to me. A drawing of a few penguins standing on an ice floe, reassuring each other that everything is fine. In the next frame, the camera begins to zoom out, and it turns out that the ice floe is drifting on an endless ocean.
It’s not the penguins who are wrong. Everything really is fine on those few square meters. The ice floe is still holding. It’s just that the world around it has changed in the meantime.
For a long time, I searched for reasons why I shouldn’t take this feeling too seriously yet. That we still have time. That technology is changing faster than actual customer needs. That the fundamentals haven’t shifted. There was truth in all of them. Yet the feeling that something essential had already shifted wouldn’t go away.
For a long time, I couldn’t find the right word for it. Dries Buytaert’s term “unbundling” helped me name what I’d been sensing for some time. Execution is increasingly becoming a commodity, while good decisions and responsibility are gaining in value.
My job is no longer about execution
I’ve been working in a strange environment for a while now. Alongside my flesh-and-blood colleagues, I’m having conversations with AI agents. When I bring multiple language models into the room so we can research, write, and analyze while debating with each other, I sometimes feel like a strange character from a Black Mirror episode. Not because I’m afraid of them (though I am, actually). But because they really do do a lot of things faster and better. They scan through material in seconds, extract the essence, and produce a synthesis that would have taken me weeks to complete.
And it doesn’t end there. If I don’t tell them what “good enough” means, they’ll keep going indefinitely. My job is no longer the execution, but knowing when it’s done. That hurts. It’s hard to get excited about something that’s rewriting the very foundation of part of my professional identity. And this isn’t just theory. It’s there in the weekly meetings, in the assignment of tasks, and in who’s willing to take on what.
Two months' worth of work in two days
Ancsa in two days into a web publication from the House of Traditions’ 430-page Folk Art Report. A couple of weeks ago, this would have taken two months to develop. Now someone just said, “Come on, we’ll do this in a few hours.”
One sentence, and the entire infrastructure we’d built up over the years suddenly seemed redundant. It wasn’t the work that disappeared. It was the framework within which we usually interpret it.
We put a lot of energy into creating a free work culture built on curiosity. The account manager works with the client to define the tasks. The developers choose among them based on their taste and curiosity. The complexity of the task is the basis for billing—almost all of our business processes are built on this. For now, for lack of a better option, this is how it remains, but we all know that we need different metrics. A different logic is taking shape, and I can’t yet see which metrics we’ll settle on.
It’s hard to describe this without making it sound like a loss. Because it is.
Then there’s the web search engine. For years, we were proud of our integrated SolR solution: it handled keywords and synonyms intelligently. It searched for words. Ancsa linked it to a language model, and from then on, it no longer searches for words, but for meaning. It’s not the same thing. Suddenly, it understands what you’re asking in Bengali and Urdu—languages we never even dared to imagine before. The penny dropped: the question isn’t whether we can do it. It’s whether we believe it can be done.
What This Is Really About
The more I delve into this topic, the less I feel like I’m thinking about AI. Rather, I’m thinking about us. We built Integral Vision on a certain worldview. One in which technological knowledge is scarce. In which development is time-consuming. In which implementation is a value in and of itself.
These assumptions didn’t change overnight, but they’re no longer as certain as they were a few years ago. For a long time, I was preoccupied with what would remain of what we had built. Now I’m increasingly interested in what can be built upon it. It seems like a small difference. Yet for me, this is what changed everything.
Share with your friends!