We run Scrum. Two-week sprints, planning, refinements, retros. And I never really questioned it — until I started noticing a pattern.
It wasn't one big moment. It was a bunch of small ones. The most recent: we were in a refinement session, talking through some relatively small change. The usual back and forth. At some point I just threw the problem into a Cursor agent while the discussion was still going — and by the time we wrapped up, there was a ready-to-review PR sitting in GitHub.
That keeps happening. And it makes me wonder what we're actually optimising for.
What Scrum Actually Gives You
The ceremonies aren't the point. The point is what they produce: a sense of what your team can deliver in two weeks, what to prioritise, and whether you're on track. That's genuinely useful. Customers expect features when you said they'd arrive. The team needs to know what matters this sprint and why. Without some structure around that, you're just reacting.
Velocity, capacity, roadmap visibility — I'm not arguing against those. They're real, and they matter more as you grow. At NetCero, we're a product-based SaaS with real customers and release expectations. That structure exists for a reason.
The Part AI Is Breaking
The problem is that most of the ceremony exists to answer one question: how long will this take?
Story points, refinement discussions, sprint capacity planning — all of it sits on the assumption that estimation is meaningful and stable. That assumption is getting harder to hold onto. When a task that used to take a day takes an hour — sometimes ten minutes — the planning overhead starts to feel completely disconnected from reality. You can't velocity-track your way through a world where execution speed is becoming unpredictable in the best possible way.
The Hypothesis
Strip everything that exists for its own sake. Keep what produces the actual outputs you need.
There's a broader shift happening. A wave of founders coming out of SpaceX and Starlink are building their own companies with the same instinct baked in from day one — rethink every process from scratch, cut what doesn't survive first principles, treat meetings as a cost not a default. I watched an interview recently with one of them and the throughline was clear: the overhead isn't neutral. It compounds. And the teams that move fastest are the ones who never let it accumulate in the first place.
Tobi Lütke at Shopify does this literally: every 18 months or so, all recurring meetings get deleted. The important ones come back. The rest don't, and nobody misses them. That's the right instinct — not as a productivity hack, but as a forcing function to figure out what your process actually needs versus what it's just accumulated over time.
We're currently in the middle of rethinking this ourselves — nothing decided yet, no changes made. But the direction is clear: keep roadmap planning, keep prioritisation, keep the big-picture alignment. Cut the refinement sessions for small tickets. Cut the ceremonies that exist to generate estimates for work AI will complete in a fraction of the time anyway.
Minimum viable process. Not no process.
Top comments (0)