TL;DR: AI sped up creative production but didn’t touch the operation around it: briefing, scheduling, handoffs, asset retrieval, review. That mismatch is the “compression test.” Teams built for survival absorb the extra speed as pressure, not capacity. Burnout shows up in PMs and team leads instead of makers, senior people become traffic controllers, and finished work sits waiting longer than it took to make. Passing the test isn’t an AI problem, it’s an operations problem, and it has to be solved before the speed arrives, not after.
AI made creative faster. So why does everything feel slower? Take the test.
Think about walking through airport security when a flight surge hits. More passengers arrive per minute. The scanners don’t multiply. Nobody in the queue is moving slower than usual. But the line grows anyway. This is because the infrastructure around the passengers was sized for a different volume, and no amount of faster walking fixes that.
That’s what’s happening inside creative organisations right now. AI made the work faster. Nothing around the work changed. I’ve now watched this same collision play out inside enough organizations we work with that I stopped treating it as isolated bad luck and started treating it as a pattern. One predictable enough to test for.
So I built a test for it. I call it the compression test: a way of determining whether an organization’s problem is AI adoption, or whether the operation around the AI was never built to absorb what it unlocks. Here’s what it is, why it matters, and how to know which side of it you’re on.
What the compression test is
AI compressed production. A concept that took a week may now take a day. An asset that took a day may take an hour. But production was only ever one part of how creative work moves through an organization and it was rarely the slowest part.
The work still has to be briefed, and most briefs are still written the way they were five years ago. It has to be scheduled, and most project management still runs on cadences built around week-long tasks, not hour-long ones. It has to move between people, and most collaboration still happens across disconnected tools, status meetings, and threads nobody can find. It has to be versioned, stored, and retrieved, and most asset management still can’t tell anyone which file is the final one. It has to be reviewed, and feedback still arrives in fragments, from too many voices, on nobody’s deadline.
AI touched none of that. The compression test is simple: when production accelerates, does the operation around it absorb the speed, or choke on it?
Why it matters more every month
Here’s the uncomfortable part: the friction was always there. It was just invisible, because production was slow enough to absorb it. When the design took a week, nobody noticed if the brief took three days, the feedback took four, and the final file lived in someone’s downloads folder. The friction fit inside the slack.
AI removed the slack. Every operational weakness that used to hide inside long production timelines is now exposed, all at once, and it compounds. Faster production means more work in motion at any given moment, which means more briefs, more handoffs, more versions, more feedback loops running simultaneously through systems that were straining at the old volume. This is why adoption numbers keep climbing while creative velocity barely moves. It’s why the work feels both faster and slower at once. It’s why teams report higher satisfaction and higher burnout in the same breath.
None of that is a paradox, and none of it is an AI problem. I know because I keep seeing the same failing result inside otherwise excellent organizations (strong teams, good tools, genuine adoption), dressed up as tool fatigue so nobody has to look at the operation underneath it. And it doesn’t fix itself. Adding more AI to a fixed constraint doesn’t create flow. It creates pressure.
How to know if you’re passing
The test doesn’t show up in a dashboard. It shows up in behavior. Three places I look first:
Is burnout appearing in new places? Burnout used to live where the production pressure lived: the designers, the writers, the makers. If it’s now showing up in the people around the work (project managers drowning in volume, reviewers with a queue that never clears, team leads spending their days routing instead of leading) that’s the signature of a failing test. The production load lightened. The coordination load exploded. And it landed on the people whose roles were never resized for it.
Are your best people becoming traffic controllers? Watch what your most senior creatives actually do all day. If the answer is chasing feedback, reconciling versions, clarifying briefs that arrived half-formed, and sitting in status meetings about work that’s already finished, the operation is spending its most expensive judgment on its cheapest problems. That’s not a workload issue. It’s speed with nowhere to go, being absorbed by seniority instead of structure.
Is finished work waiting longer than it took to make? This is the one to actually measure. When an asset takes an hour to produce and three days to get reviewed, approved, found again, and used, the operation is now the long pole in every timeline. Most organizations have never tracked how long work sits still, because it never used to matter. It’s now the single number that decides whether AI is making them faster or just making them busier.
How we ended up on the right side of it
We didn’t build for AI. We built years ago to solve a completely different problem: creative operations that lurched between idle and drowning. The fix was never about the creative work itself. It was about everything around it: how work gets briefed, tracked, moved, stored, and closed. A small permanent core with external capacity, and underneath that, an operational layer designed so work never waits on the machinery around it.
As Walter, our Head of Creative, puts it:
“We’re seeing exponentially more throughput since bringing AI properly into our workflow — but the throughput was never the hard part. The hard part was done years ago: finding every point where work sat still. Briefs that took longer to write than the work took to produce. Plans built around weekly check-ins when the work was moving daily. Feedback arriving in five formats from five directions. Assets that took longer to locate than to create. We went after all of it — not because AI was coming, but because work sitting still is the most expensive thing in a creative operation, at any speed. When AI arrived, there was simply nothing left for the speed to get stuck on.”
That’s the part worth sitting with. Passing the compression test was never an AI project. It was an operations project that happened to be finished before the test began.
The responsible growth question
This is, in the end, the same principle we’ve been writing about all year. There’s a version of growth that adds (more tools, more speed, more output pushed into a system that was already straining) and calls the resulting pressure progress. And there’s a version that builds the operation first, so that every gain lands as capacity instead of chaos.
AI has made the difference between those two impossible to hide. The organizations bolting speed onto an unchanged operation aren’t growing, they’re accumulating pressure and spending their people to absorb it. The organizations that fixed the operation first are now compounding every gain the technology hands them, without the burnout bill.
That’s the compression test. Everyone’s taking it. The only question left is which result you’re currently producing, and whether you’ll fix the operation before the next wave of speed arrives, or after it.