What it looks like
The velocity charts are fine. The roadmap is full. Nobody is slacking, and there is no obvious villain. But the needle the team exists to move has not moved in quarters, and the unspoken response has been to add process: more discovery templates, more prioritisation frameworks, more alignment meetings. Each addition feels rigorous. Collectively they are sediment.
What is actually happening
The team has hit the ceiling of its operating model, not the ceiling of its talent. Three flows determine what a product team can do: how work flows (queues, handoffs, batch sizes), how decisions flow (who can say yes, and how long that takes), and how learning flows (how fast a signal from a user changes what gets built).
When effort stops translating into outcomes, one of these three is silted up, and pushing harder increases the pressure without increasing the throughput. The cruel part is that the visible response, working harder and adding process, actively worsens the constraint, because every new artefact adds another queue.
AI sharpens this pattern rather than softening it: it collapses the cost of building and experimenting, which means the constraint shifts even harder onto deciding and learning. A team that could not prioritise at the old speed of building definitely cannot prioritise at the new one.
The intervention shape
Map the three flows against real recent work and find which one is binding. Then remove sediment rather than adding process: fewer queues, shorter decision paths, faster signal loops. The fix is usually subtractive, which is why teams rarely find it themselves; nobody gets promoted for deleting process.