The Observation
Every few decades, a technology removes a constraint that organizations had built their entire operating model around. Resistance to change was once a function of how long it took to retool a factory. Then of how long it took to ship software. Now, AI agents are removing the execution constraint itself.
When execution accelerates by an order of magnitude, the bottleneck shifts. What remains is the decision: what to build, for whom, and why. Who has the authority? Who has the context? Who is willing to be wrong?
These questions are not new. In manufacturing, logistics, and finance, organizations have wrestled with decision-making speed for decades. What is new is that software organizations — long immunized by talent scarcity and a “part of the solution” self-image — are confronting them for the first time. The special economic conditions that shielded software teams are ending. AI breaks the homogeneity. The bottleneck becomes visible.
In software, the symptoms are vivid: a single developer with an AI agent completes in one day what once took a team two weeks. Backlogs empty. Sprint reviews drown in hundreds of stories no one can absorb. The universal reaction is to abolish Scrum — remove the process, go faster.
This is the wrong response. Not because Scrum was right, but because removing it hides the real problem instead of solving it.
The real problem was never execution speed. It was always the decisions. AI agents removed the comfortable fiction that if only we were faster, everything would work. The water level dropped. The rocks are visible now.
The Name
In 1978, Herbert Simon received the Nobel Prize for a deceptively simple insight: organizations are not systems of people — they are systems of decisions. Every structure, every hierarchy, every process exists to channel decisions. Remove everything else, and decisions are what remain.
In 1971, biophysicist Manfred Eigen described how simple chemical systems organize themselves into self-sustaining cycles — hypercycles — that enable the leap from chemistry to biology. From mechanical to alive. From simple repetition to emergent complexity.
We believe organizations face a similar transition. From process-driven to decision-driven. From coordinated to self-organizing.
We call this transition hycle — the hyper cycle.
German speakers hear heikel — delicate, precarious. That is deliberate. This transition is not comfortable. It challenges power structures, role definitions, and the identity of everyone involved. It is, indeed, heikel.
And yes — hype cycle is in there too. Deliberately. Every few years, a new technology triggers the same predictable curve of inflated expectations, disillusionment, and belated productivity. We named ourselves after that pattern as a reminder: hycle is not about riding the next wave. It is about what remains after the hype has passed.
What We See
These observations come from practice — from training teams, consulting organizations, and building software with AI agents ourselves.
1 AI agents shift the bottleneck.
Execution speed is no longer the constraint. Decision-making is. In software, one person builds an epic per day. In other domains, AI compresses analysis, drafting, and production to a fraction of their former duration. In all cases, the organization cannot decide what to do next at the pace the work now demands.
2 The decision problem is not new — it is newly visible.
Like lowering the water level in a river reveals the rocks beneath, AI speed exposes organizational dysfunction that was always there. Slow decisions, unclear authority, missing domain knowledge, absent stakeholders. Many industries outside software have known this for decades. The difference is that software organizations are now losing the immunity that talent scarcity and long delivery cycles once provided.
3 Process models designed for throughput optimization break down.
Scrum, SAFe, and similar frameworks optimized for a bottleneck that no longer exists. Sprint planning for 300 stories is not planning. Reviews of work already in production are not reviews. But not everything in these models was throughput optimization. The cadence — the regular rhythm of reflection, inspection, and adaptation — retains its value. What collapses is the scaffolding around delivery speed. What remains worth preserving is the discipline of stopping to think.
4 Teams shrink. Responsibility grows.
One- to two-person teams with AI agents carry more domain scope than eight-person cross-functional teams did before. But the organizational structures around them have not adapted.
5 Role boundaries dissolve.
The separation of “developer” and “product manager” loses meaning when execution is instant and the bottleneck is domain understanding. We see these roles converging — toward people who combine deep domain knowledge with the ability to deliver end-to-end and make decisions autonomously. We are not ready to name this role. But the convergence is unmistakable.
6 The coordination layer faces an existential question.
Middle management justified its existence through managing development complexity. That complexity is evaporating. But the challenge extends beyond middle management. Boards, founders, and executives face their own reckoning: when teams can move faster than the hierarchy can direct them, what is the role of leadership?
7 Organizations regress under pressure.
In times of high uncertainty, leaders reach for central control — not because the situation demands it, but because the need to feel effective disguises itself as the need to lead. Organizations that had developed toward distributed decision-making get actively reset to command-and-control by their own founders. The drivers are identity, role confusion, and unexamined certainty — not evidence. Progression through the maturity levels is not monotonic.
What We Believe
We do not have answers. We have hypotheses. They come from practice, not theory. We hold them loosely and test them rigorously.
H1 Organizations must reorganize around decisions, not around execution processes.
The fundamental unit of organizational value is a decision — not a feature, not a story, not a sprint goal, not a deliverable.
H2 Organizational maturity is determined by where decisions are made — not by how fast work is shipped.
Speed without decision authority produces faster waste.
H3 Autonomy without sharpness produces faster ambiguity.
Distributing decision authority solves nothing if the decisions themselves remain diluted. Real decisions require contrast: genuine alternatives, a real trade-off, and someone willing to stand behind the consequence.
H4 Decisions that cross team boundaries need explicit architecture.
A decision must be owned at the level where it first affects more than one entity. Without this principle, team-level autonomy creates local optimization at the cost of organizational coherence.
H5 Frustration is a signal, not a problem.
The discomfort teams feel when AI speed meets organizational inertia is the indicator of where to intervene next. Do not hide it. Follow it.
H6 The path is iterative and bottom-up.
There is no master plan for this transition. You navigate from bottleneck to bottleneck. Each one reveals the next.
H7 The default shifts from instruction to intent to autonomy.
Autonomy — for teams and for AI agents — must be earned through consistent delivery, not declared by management. You cannot skip levels. There are no shortcuts.
H8 The opportunity is growth, not reduction.
AI that accelerates execution unlocks a choice: do the same with fewer people, or do fundamentally more with the same people. The first is a one-time saving. The second is a competitive advantage that compounds.
The Model
The hycle model describes organizational maturity along two co-evolving dimensions:
These dimensions are not independent. You cannot advance on one without advancing on the other. An organization that grants AI agents full autonomy without team-level decision authority will build the wrong thing faster. A team with full decision authority but no trust in their AI tools will be too slow to use it.
The transition between levels is not linear. Organizations do not “implement Level 3.” They earn it — through practice, through pain, through honest confrontation with their own decision-making structures. You cannot skip levels. There are no shortcuts. And progression is not guaranteed — under pressure, organizations routinely regress to earlier levels. The pull toward central control is strong, especially when uncertainty is high.
You are here if: Teams complain about having nothing to do.
You are here if: Frustration points at the organization, not at the tools.
You are here if: Teams ask “What should we build?” more often than “How should we build it?”
You are here if: Decisions emerge from customer need, not from hierarchy.
You are here if: We do not know yet. This is where we need your experience.
What does it take to move from one level to the next?
We have early hypotheses. We are testing them with real organizations. This is the question we most need help answering.
What hycle Is Not
hycle is not a new Scrum. It is not a process framework with ceremonies, roles, and artifacts to implement on Monday morning.
hycle is not against Agile. The values of the Agile Manifesto still hold. The process models built on top of them do not — not at this speed.
hycle is not a technology framework. It does not prescribe tools, architectures, or AI platforms.
hycle is not finished. It is a set of observations and hypotheses from practice, offered as an invitation to collective sense-making.
The Roles
When execution is no longer the bottleneck, the division between “the person who decides what to build” and “the person who builds it” stops making sense. We see these roles converging — toward people who combine domain expertise, discovery capability, and end-to-end delivery in a single responsibility.
We are not ready to name this role or define its boundaries. But the convergence is real, it is accelerating, and it changes what organizations need from the people inside them.
Coming soon.
The Language
One of the most underestimated contributions of Scrum was not its process — it was its vocabulary. Sprint, backlog, retrospective, product owner: these terms gave organizations a shared language for coordinating work. That language is losing its referents. The concepts it describes are dissolving.
If hycle describes a transition from process-driven to decision-driven organizations, that transition needs its own shared language. Not jargon. Not a glossary. A way for people across roles and levels to talk about decisions, authority, trust, and autonomy with precision and without ambiguity.
We are working on this.
Coming soon.
Walk the Talk
We apply these ideas in our own work and with our clients. This section will grow as we document what works, what fails, and what surprises us.
Coming soon.
The Authors
hycle emerged from a conversation between five practitioners who have spent decades building, consulting, and transforming organizations.
The Invitation
hycle is not a declaration. It is an invitation to a conversation.
We believe the organizations that thrive in the age of AI agents will be the ones that learn to make better decisions faster — not the ones that execute more work faster.
If you see what we see. If you have observations that confirm or contradict our hypotheses. If you are navigating this transition in your own organization — we want to hear from you.