So we built something else. A distributed collective of senior people: remote-first, documentation-driven, async where it makes sense, assembled around the problem instead of around a payroll we had to keep busy. At the time that looked like a lifestyle choice. It wasn't. It was an operating model, and it's the one the market is now moving toward.

AI is speeding that up.

Most of what gets said about AI and work is either apocalyptic or empty. It won't replace everyone and it won't solve everything. What it actually does is less dramatic and more uncomfortable: it makes inefficiency visible.

Clients have felt it for years: that projects drag, that too many people are involved, that half the timeline is coordination rather than work, that some processes exist mainly to justify the size of the company running them. Traditional agencies aren't in trouble because AI can draft a layout. They're in trouble because AI strips away the economic cover that protected slow, heavy ways of working.

Once research, prototyping, documentation and a lot of production can be done in a fraction of the time, the client's question answers itself. Why keep paying for the big team, the management layer and the rigid process when a small senior team moves faster and you talk directly to the people doing the work? The pressure isn't really coming from the technology. It's coming from clients who finally have a better option.

For a long time, size sold. A big agency could promise capacity, specialists, backup, a nice office, a client list that looked reassuring in a pitch. Serious work meant a serious-looking agency, and serious usually meant large.

In digital work, scale cuts the other way. Big structures have to be fed. Fixed teams have to be billed out. Hierarchies need meetings and reporting lines, junior-heavy delivery needs senior review, and every department adds a handover. Every handover adds friction, and friction is just cost with better manners.

Clients used to accept this because there was no real alternative. There is now, and they understand the work far better than they did ten years ago. They can tell when a team is too big or a process is heavier than the job needs. A handful of experienced people can do in days what used to take weeks of back-and-forth. That doesn't reduce the value of expertise. When the tools get more powerful, judgement matters more, not less.

This is the belief we've worked from since the start: the best results don't come from putting as many people as possible on a project. They come from putting the right people on it, at the right moment.

It sounds obvious, and it quietly contradicts how most agencies are built. Work moves down a line, from strategy to UX to design to development, with project managers holding it together while juniors execute and seniors supervise. It can work. It's usually slow, expensive, and a little disconnected from the actual problem.

We'd rather have a small team of senior people who hold the whole context and can think, decide and build without hiding behind process. Specialists come in because the project needs them, not because a department needs the hours. Our collective is spread across Switzerland, Europe and beyond, employed and freelance, which is what lets us work this way. The flexibility isn't a perk. It's the thing that keeps us from turning a client's budget into a solution for our own utilisation problem.

Old agency economics run on time: more people, more hours, more billable capacity. That creates a strange incentive, where efficiency costs you money and complexity earns it. It's a bad deal for the client, and most clients now know it. Nobody hires an agency because they want ten people on a status call. They want someone to understand the problem, challenge the assumptions, set a direction and turn it into something that works.

We're open about using AI, not as a talking point but as normal modern practice. If a task can be done faster, it should be, and the client should feel that in the timeline and the invoice. The point isn't to take human thinking out of the work. It's to spend it where it counts: strategy, architecture, design decisions, the specific problem in front of us, rather than on repetitive production.

The honest part. It would be lazy to say juniors don't matter; every industry needs a way to bring new people up, and ours is no different. But the junior-heavy model is under real pressure. It relied on young people carrying the production load while seniors reviewed and presented, and a lot of that production layer is exactly what AI now accelerates.

So the way into the industry has to change. Learning has to be more deliberate and closer to strategy and real problems earlier, because sitting people on repetitive work won't teach them enough anymore. And clients will rightly ask why they're paying senior rates for junior work that then needs senior correction.

Traditional project management grew up to manage complexity: the more people and dependencies, the more coordination you need. Project managers became the glue holding an inefficient structure together.

It doesn't disappear in a leaner model, but it changes shape: less chasing and status theatre, more consulting, systems thinking and decision-making. Good documentation and async workflows already remove a lot of meetings for us, and a senior team simply needs fewer handovers. The best project management isn't managing chaos well but not creating the chaos in the first place.

Remote work used to be framed as culture: some people liked it, some thought it was less serious. That's a dated argument. Done properly, remote-first is an operational advantage: it forces clearer documentation, sharper responsibilities, better async thinking, and access to talent you'd never reach through local hiring.

It isn't automatically good. Badly run remote work is as painful as any bad office. But with senior people, real systems and a culture of ownership, it beats the office-centric model. We've worked this way since 2013. We keep a small headquarters in Zurich (for now); we've just never confused office size with capability.

Plenty of agencies still assume their past protects them: the client list, the awards, the name. Those things matter. Trust and reputation are real. But a famous agency that's slow and expensive won't stay attractive on reputation alone, not when the gap between what a lean expert team delivers and what a large agency costs keeps getting more visible.

The future won't reward agencies for being large. It'll reward them for being useful: fast without being careless, strategic without being vague, fluent with the technology but still recognisably human.

If you're choosing a digital partner, the old questions don't cover it. Don't just ask how many people work there; ask who will actually be on your project. Don't just ask where they're based; ask how they get to the right expertise. Don't just ask about process; ask which parts of it create value and which just create cost. Don't just ask whether they use AI; ask how, how openly, and whether the savings reach you.

The partner you pick shouldn't make your project heavier. It should make it clearer.

Anja and I founded PAKD in 2013 as a home for good creative and technical work, and it's grown into something more deliberate without losing that. But the part that matters now, is how we built it: a flexible senior collective, remote-first, documentation-driven, transparent about our tools, assembled around the actual need.

We're not reinventing ourselves as a leaner, distributed, AI-assisted agency because the market shifted. We started there. We don't claim to have it all worked out. The tools are moving fast, and everyone is still learning what responsible, genuinely useful AI-assisted work looks like. But we're confident about the shape of it. The next agency won't be defined by headcount, office size or legacy. It'll be defined by expertise, adaptability, transparency, and the ability to turn complexity into progress. It's the model we've been building from day one.