The refinement session is running. A story is on the board. The team is looking at the acceptance criteria, estimating the work, asking what done actually means for this one.
Engineering pushes back. The scope is broader than the story suggests. The acceptance criteria miss edge cases entirely.
Product acknowledges it and redirects. There are commitments. There is a timeline. The narrative stays on delivery. The story gets pointed, the Definition of Done gets written, and the meeting moves on.
Nobody overruled engineering. The narrative just never made room for what they were saying. The sprint moves forward. The story carries a Definition of Done that reflects what the pressure allowed, not what complete actually requires.
Most organizations look at that moment and see a team dynamic. A communication problem. Something to address in the retrospective.
It isn’t a team problem.
The problem isn’t the tradeoff. Teams make deliberate decisions to deliver a smaller slice now and build the rest later. That’s legitimate. What makes it dysfunction is the silence around it — when nobody with authority names what’s being traded away, nobody tracks what was left behind, and the gap between what was promised and what was built gets quietly absorbed into the next sprint. Phased delivery decided out loud is a strategy. The same outcome arrived at through unspoken pressure is a quality decision made by default, with no one’s name on it.
Senior management sets priorities and protects timelines. Product builds the roadmap inside those constraints and owns what ships and when. The Scrum team writes the Definition of Done — what complete looks like for this epic, this story, this piece of work. The pressure that shapes what gets written into that document travels the whole length of that chain before it arrives in the refinement session. By the time it reaches the room it has no return address.
Product’s lane is to define what the work needs to accomplish from a business outcome perspective. What complete looks like. Not how it gets built. But Product is writing those definitions inside pressure that arrived before the work started.
The pressure doesn’t announce itself. It arrives as stakeholder urgency. Client requests. Market timing. Initiatives that need to move. Product isn’t misrepresenting it — those things are real. But the volume, the speed, and the answer always being now — everyone in the room knows where that’s coming from. No one says it out loud.
This is where the lane shifts. Under sustained delivery pressure, Product stops defining what done looks like and starts describing how the problem should be solved. The timeline is fixed. The scope is protected. Something has to give. Engineering loses authorship of the how. The Definition of Done on the story reflects the anxiety of a protected roadmap, not a genuine quality standard.
That happens on one story. Then the next. Then the next.
Engineering pushes back each time. Each time the narrative holds. And after enough cycles, the pushback gets shorter. Not because engineering stopped seeing the problem. Because they already know how the conversation ends.
Nobody said lower the bar. The bar got written by people answering the wrong question under pressure that arrived without a name.
Leadership looks at quality problems and sees a team failure. A process gap. Something for the Scrum Master to address. They don’t see their own decisions in the DoD because those decisions never arrived as decisions. They arrived as urgency. As priorities. As things that needed to happen now, translated through legitimate business language and delivered without a return address.
The Definition of Done on every story is a precise record of what the organization actually made possible when the pressure was on. Not what leadership said they valued. What the conditions they created made room for.
That document doesn’t lie. Most organizations just don’t know how to read it.
Now add AI to that system.
Agents don’t sit in refinement. They don’t push back on scope. They ship to the standard they’re given — faster and at higher volume than any engineering team could.
The friction that used to slow bad standards down disappears. The gap between what leadership thinks is being built and what is actually being built widens — faster than before, at greater scale, with less signal that anything is wrong until something breaks.
The Definition of Done was always a leadership problem dressed as a team artifact. In an AI-assisted delivery environment, the cost of leaving it that way just went up.
Want the Experiment-Driven Agile Retrospective Toolkit?
If you’d like the Toolkit, reach out and I’ll send details (what’s included, pricing, and how teams use it). Or subscribe for new posts and updates.