Engineering Beyond Agile: When teams become cells
How Context Replaces Coordination
Until recently, I explained most of this with diagrams.
AI‑native SDLC. Agentic DevOps. Value streams with feedback loops, control points, and stages that made sense because we’ve been using that visual language for decades. Boxes and arrows showing how work moves, where it slows down, where it needs guardrails.
Those diagrams were useful, and to some extent they still help explain what exists and how things connect. But over time, I noticed they were no longer answering the questions engineering leaders were actually asking.
They describe flow and activity, but not behaviour or tension. They also assume the system is still organised around handoffs and stages, even as decisions surface and collide in different ways.
What those diagrams don’t capture is how organisations experience execution once it accelerates. They don’t show how context fragments, how disagreement appears earlier, or how teams start interacting before intent has fully settled.
How time used to absorb uncertainty
For a long time, organisational design relied on time to do a lot of quiet work.
Work moved through stages. Business intent became product direction, product direction became engineering work, engineering work became operational reality. Each step provided space to interpret, negotiate, and defer decisions. Disagreement could remain implicit, and assumptions could stay unstated until implementation forced them into the open.
That worked because execution took long enough for meaning to be worked out along the way.
As execution accelerates, that buffer disappears.
Decisions no longer wait for their designated phase. They surface wherever intent is ambiguous, constraints are incomplete, or trade‑offs haven’t been made explicit. Often in parallel. Often earlier than teams are used to handling them.
The organisation may still be structured like a pipeline, but it no longer behaves like one.
When teams stop behaving like stages
In this environment, teams begin to operate differently, even if nothing on the org chart has changed.
They stop functioning as stages that receive work, transform it, and pass it on. Instead, they behave more like cells in a system: locally accountable, partially informed, and tightly coupled through shared decisions.
Each team carries a piece of intent. None of them holds the full picture. And none of them can make decisions safely without some shared understanding of what the others are optimising for.
Coordination doesn’t disappear. It becomes more distributed, and more sensitive to gaps in shared understanding.
Context becomes the thing that has to travel
In pipeline models, coordination is managed through plans, meetings, and handoffs. Context is exchanged verbally or through tribal knowledge, and gradually converges over time.
In faster systems, that approach breaks down.
Context has to move at the same speed as execution. It has to carry intent, constraints, assumptions, and trade‑offs across teams without being renegotiated from scratch each time.
This is why teams often feel less confident even as they become more productive. It isn’t because they’re moving too fast. It’s because the mechanisms that used to carry context can’t keep up.
Where alignment actually lives
One of the quieter shifts teams encounter is that alignment no longer lives primarily in meetings. They still matter, but they stop being the place where alignment is completed. They become places where signal is generated: questions, tensions, unresolved trade‑offs.
If alignment depends on those conversations alone, it decays as soon as execution accelerates.
This pushes alignment into artifacts that persist beyond the room. Not as documentation for its own sake, but as places where intent and constraints are made explicit enough to support fast decisions.
This shift is often uncomfortable. It changes who has influence. When alignment lives in meetings, influence flows to those who can negotiate meaning in real time. When it lives elsewhere, influence flows to those who can clarify intent precisely and make assumptions visible.
That discomfort is often attributed to AI. It is more accurately about where authority is exercised.
Collaboration without handoffs
As teams start behaving like cells, collaboration stops looking like handoffs and starts looking like shared context.
Product teams, platform teams, business teams, and increasingly agentic capabilities interact through overlapping context. Decisions propagate through shared understanding rather than formal approvals.
This is also where many early agentic experiments struggle. Agents accelerate execution, but they don’t resolve ambiguity. Without shared context, they amplify divergence. With it, they reduce the cost of iteration.
Execution can be delegated, yet responsibility for context cannot.
The role of enablement and expertise
As decisions move earlier, traditional governance mechanisms struggle to keep up. Late reviews, approval boards, and policy decks (yes, that) were designed for slower systems. They become visible, but ineffective. Present, but unable to shape decisions before they harden.
In organisations that adapt, enablement teams and SMEs don’t sit at the end of the flow. They shape the environment the system operates in. They make constraints explicit. They encode shared understanding about risk, quality, and acceptable trade‑offs. Their influence surrounds the system rather than intervening in individual decisions.
This is how trust scales when manual oversight no longer does.
The elephant in the room org: conflict
Decisions that used to surface during implementation, integration, or post‑incident review now appear while intent is still being formed. That can feel more intense. More personal. And sometimes more political.
It isn’t. The timing has changed.
When time no longer absorbs uncertainty, disagreement has to surface somewhere. Organisations that aren’t designed to handle it early experience it as friction. This is why a growth mindset culture is the important part of the DNA.
What changes when teams become cells
When we stop thinking in flows and charts, and teams become cells, collaboration stops being about coordination and starts being about maintaining shared context. I learned to understand why talking about removing walls/silos isn’t resonating looking at an org chart. It’s a misrepresentation of how we can do future engineering. The work, accountability and decision making moves earlier. is shared. It is exactly what makes speed sustainable.
In the next article, I’ll look at what happens when these early decisions involve irreversible trade‑offs like security, cost, and capacity.
This is the fourth article in this series “Engineering Beyond Agile”. Read my previously published articles.