Ana Marković
Building got cheap and generating got free. What got scarce is a room full of capable people landing, together, on what to build and why.
Everything is buildable now. Every team I meet lives in the same climate: more ideas than quarters to test them in, more tools than problems, roadmaps that read like wish lists because nothing on them is technically impossible anymore. Ideas were never scarce, and lately neither is making. What got scarce is convergence - a room full of capable people actually landing, together, on what to build, for whom, and why this and not the ten-thousand other things.
Three acts follow, as attempts to buy that back - and they rise the way acts should, toward the one part no method can reach.
A client of ours - let's call them Zenith for the purpose of this blog - started as a sketch. A founder arrived with just an idea, and it was up to us to lay down the steps we'd have to march to bring it to life.
A design sprint is exactly what the name promises: a deliberately short distance, run flat out, shoulder to shoulder. The distance is the trick - a team can only run all-out when it can see the finish line from the start. And the shoulders matter just as much, because despite the first word of its name, this was never a designer-only sport. The room holds the whole company in miniature: engineers, founders, designers - whoever owns the problem - and nobody gets to spectate. Early on, anything goes - rough sketches from engineers, doubts from the founder, the idea no one would risk in a normal meeting. Teams rarely run out of ideas, they run out of decisions.
So the course is staked with them. At marked points the group chooses, at speed: sticker votes, dots piling up, one direction taken, the rest left behind on the record and without a fight. And what waits at the finish line is not a report or a slide deck. It is a prototype - a film set of a product - put in front of real users, so that reality gets a vote while being wrong is still cheap. That is the sprint's real trade: the momentum of a hackathon aimed at the one thing hackathons never produce - a decision the whole team believes in, because the whole team ran it down.
Zenith is live today. It has paying users. And underneath all of it - the product, the revenue, the users - the first artifact in the chain is still a pencil sketch.
The sprint's success creates its own temptation, which is to declare it. The Answer - run the same framework for everyone, laminate the agenda, retire the thinking. But a sprint assumes a particular shape of problem, and clients don't arrive in one shape. Some arrive the way Zenith did - a founder, a drawing, a decision one good sprint away. Some arrive as a scale-up with three roadmaps and, on a good day, one company. Some arrive as an enterprise that doesn't need another prototype at all; it needs forty people to mean the same thing when they say the word "priority". These clients are not the same distance from a decision - they aren't even running toward the same finish line. Run them all through the identical agenda anyway and you haven't been rigorous. You've been convenient. Rigor is redesigning the race for the runner.
So we did the most self-aware thing a design team can do: we got in a room in Lisbon and ran a workshop about workshops. The goal was practical: get everything we had learned from every client out of our heads, find the patterns in it, and turn those patterns into new workshop frameworks - so the next client who walked in wouldn't get the default agenda. They'd get one built for them.
A workshop, done right, is an orchestrated brain dump. The dump is not optional - people walk in already full: opinions, grievances, half-formed ideas, the roadmap they'd write if anyone asked. The only question is whether all of it lands into a structure or onto the floor. Orchestration is the difference: sequence, roles, time boxes, and a person whose job is friction. In Lisbon, the dump was every client type we'd actually met, every issue they'd brought versus the one they'd booked us for - and the structure it landed into was a map: client types against their real problems, problems against exercises, exercises against the scenarios where they earn their keep.
What has vanished from a culture of endless, frictionless production is the "no". Everything about how we work is optimized toward "yes": tools that autocomplete, roadmaps that only grow, meetings engineered for agreement, machines that will generate a hundred versions of anything before anyone asks whether there should be one.
We built ourselves a yes-machine, and the "no" lost its habitat. Designers have an old name for what went missing: negative space - the part you deliberately leave empty, which is the part that gives the whole thing its shape.
A workshop is one of the few rooms left where the "no" is on the agenda. Some of it happens at the composing desk, map open - which exercises to run, which to leave in the drawer. The harder one lives in the room, and the facilitator's real job is not to say it, but to help the room find it. The "no" is almost always already there. It sits buried under politeness and momentum. So the job looks like interruptions - the timer, the agenda, "let's park that" - but it works like excavation: build the exact moment where the buried thing becomes sayable, and let someone in the room say it. A "no" found by the team lands differently than a "no" delivered from the front of the room. It stops being an objection and becomes a decision - a cut, the small kind, made at minute three so it doesn't have to happen at month three.
Which brings us to the newest thing on the table, the one arriving with the most hype - which is exactly why it needs the coldest introduction.
The yes-machine has produced one brand-new, very specific epidemic. Everyone has AI now - it's in the roadmap, the board deck, the job titles - and almost nobody can say which of their problems it actually maps to. The capability arrived before the use case. Teams don't lack access to intelligence; they lack a structured way to figure out where it belongs - which corner of the workflow, which user pain, which product surface. There is, genuinely, no established method for that conversation yet. Left unstructured, it produces either paralysis or a chatbot. Usually a chatbot.
So we went analog on purpose: a deck of cards.
The AI Cards exist for exactly that missing conversation. The deck walks a room through the mapping the hype keeps skipping - on one side, the areas where AI can actually act; on the other, your problems, your users, your product surfaces.
The joke writes itself: the antidote to infinite compute turned out to be printed card stock. But the analog is doing real work here. Cards make the option space finite and physical - you can hold a possibility, weigh it against another, and, critically, put it down. They also change the room's posture: the session plays like a game and works like a method, a conversation starter with an agenda hidden in the shuffle. The truest description I have is that the deck is a prompt, pointed the other way. We've spent the last few years learning how to prompt the machines. This one prompts the humans - it takes "we should do something with AI," the emptiest sentence in modern product work, and forces it through problems, users, and surfaces until what comes out the other side is a use case you could actually build, or the honest discovery that you don't have one yet. Both outputs are wins. Only one of them used to cost a quarter to find out the hard way.
I'd love to end with the reassuring version. The one where the methodologies solve it, facilitation is the answer, we've all found the new job and it's safe and pays well and comes with dental.
The methods help you decide. They don't hand you conviction. You can run the cleanest process of your career and still end up standing in front of four good directions with the actual choice sitting there unmade. Because the choice was never going to come from the process. The process clears the ground. Somebody still plants the flag, and the flag is heavy, and it's yours.
That's the part that didn't get cheaper. Building got cheap. Generating got free. The cut - this one, and I'll own it when I'm wrong - costs exactly what it always did. The methods are an antidote to the noise. There is no antidote to the cut.
We just built machines good enough at everything else that we can finally see it.