The Stand-up Isn’t Going Away. The Room Is
The twist of going fully AI native

Many organisations hear AI-native development and conclude that the stand-up, and most other agile ceremonies, are on their way out. The agents do the building. The spec holds the truth. A small mob has everyone it needs. Why meet?
For some teams, that is true. For most organisations, it is the opposite. AI does not remove the need to work together. It shortens the time we can afford to wait. When the economics of execution change, the economics of coordination change with them.
In “The Spec Is Not the Handover“, I argued that the daily stand-up is how product gets back into the room. This article follows that thought further.
A blocker, found the same day
In one of our recent bolts, a short and focused AI-DLC cycle, the work stopped. The AI had raised questions nobody in the bolt could answer, and the answers sat with product, in a different calendar and a different focus.
We had kept our stand-ups. So the blocker surfaced the very same day. The right people heard it, we saw what was stopping the flow, and we made the decisions from there.
Without that touchpoint, the bolt would likely have waited for days before anyone noticed. With AI, that is a long time. The build itself might only take hours. When building gets that fast, waiting becomes the most expensive part of the work.
When the ceremonies really can go
It is worth being honest about where the belief comes from.
If your whole product is one mob, two or three people working on one thing together, then everyone who holds the intent is already in the room. You communicate with those involved, and that is it. Heavy ceremonies would only add noise.
Many organisations can also move closer to this. With AI-DLC, a Kanban flow where small mobs of two or more people swarm one item at a time becomes a natural way of working. Work flows, blockers surface when they appear, and the ceremony load can be light.
So there is no single right answer. Being truly agile means fitting the process to the reality of the organisation, not copying a framework by the book.
The twist of going fully AI native
Now follow the idea to its end. If you want to be truly AI native, everything becomes spec-driven, including the small work items. Living specs that result in code.
A small spec-driven item can be a one-person job. In the language of bolts and mobs, that is a one-person mob. And when many one-person mobs run side by side inside a sprint or a Kanban flow, you are back to something very familiar: individuals working in parallel, each holding a piece of the picture. Only faster.
I am stretching the word on purpose. AI makes individual execution more powerful, and that can make collective understanding more fragile. The coordination problem has not disappeared. It has come back faster.
Why larger organisations need the ceremonies more
Teams working on larger applications still need to produce multiple things in parallel. Many will still run Scrum, or a scaled agile approach. And for them, the ceremonies do not lose their purpose. They gain urgency.
They do two jobs that AI cannot do on its own.
They keep tacit knowledge at the front of the room. In AI-native development, the AI asks the questions, and tacit knowledge is what answers them: why something matters, what the customer really meant, which part of the system tends to break. Each answer is captured in the living spec, so the spec grows with what the people know. A ceremony is where that knowledge comes forward, and where new knowledge forms, before it fades. As I wrote in “The Moat Is Moving“, tacit knowledge is a flow, not a stock.
Google’s Elephant-Goldfish Model makes this visible. The elephant holds the project context and drafts the spec. The goldfish reads it with no context at all, and probes for the gaps and unstated assumptions. In other words, the goldfish is designed to find exactly what nobody wrote down. The gaps go back to the elephant to be resolved. But the elephant only knows what has been written down and discussed. AI can expose an assumption. It cannot tell whether that assumption reflects the actual intent. That knowledge sits with people, and that is why the room matters more, not less.
They resolve bottlenecks fast enough to keep up. In the past, we accepted waiting weeks for decisions. Agile pushed us to shorten the feedback loops to days. AI-native work needs them shorter still. What used to be days is now hours.
Present, but not in the room
There is a quieter reason the room matters. Even when people sit together, AI creates wait time. While the agents work, attention drifts to other things. People are physically present, but their attention is somewhere else.
In the 1970s, Thomas Allen at MIT showed that communication between engineers drops sharply with physical distance. Shared attention research adds the other half: two people can sit a metre apart, and if their attention is not on the same thing, the distance reappears.
Being present is not the same as sharing understanding. Shared attention creates collective sense-making, and collective sense-making is what makes decisions fast. This is why the stand-up is not overhead. It is one of the few moments where attention is shared by design. It brings the room back together, for a few minutes, around the same intent.
I have written about the habit itself in “The 15-Minute Daily Habit That Connects Every Team to a Shared Truth“, and about noticing where attention goes in “How the Best Leaders Read the Room“.
Fit the ceremonies to the shape of your work
The question is not whether to keep your ceremonies. It is which shape your AI-DLC takes, and what that shape needs.
- A standalone mob of two or three: keep it light. Talk when you need to.
- Kanban with small mobs: a short daily touchpoint to surface blockers and dependencies the moment they appear.
- Many parallel bolts on a larger application: keep the stand-up and the review, and tighten them. The review is a learning system, not a demo, as I wrote in “The Sprint Review Is Not a Demo Meeting“. They are how intent travels and how decisions keep pace.
For sponsors and executives
If you are investing in AI-DLC to get 10x, the ceremonies are part of how you protect that investment. The agents can only build as fast as the decisions behind them. A team that waits three days for an answer loses more now than before, because everything else got faster.
Try this in your next stand-up
Ask one question:
What is the AI asking that nobody here can answer?
Then notice who should have been in the room, and how long that answer would have taken without the meeting.
Then measure it. Track decision latency: the time from the moment a question is raised to the moment someone answers it. In many organisations this used to be days, sometimes weeks. In an AI-native pipeline, aim for hours. If the number grows, the bottleneck is not the agents. It is the room.
AI is making execution cheaper and faster. It is also making the cost of waiting more visible. The stand-up is not going away. The room simply has less time to catch up with the work.

