Agentic Experience Optimization
Manifesto
Seven principles for the age of agents. Artificial intelligence has stopped merely responding. It has begun to act. This document sets out the principles of a discipline born from that shift: Agentic Experience Optimization.
I
We are no longer the only ones who act.
Throughout history, acting upon the world has been a human prerogative, and tools were its extension. This is no longer true. Artificial agents do not execute commands; they pursue ends. They reason, choose, remember, correct themselves. We are inside a transition that, with the emergence of new actors, redefines the human and its way of being in reality. What changes is not the power of the tools, but who performs the action. Every digital system must be rethought starting from this fact.
II
Agency is distributed, not surrendered.
That an agent can act in our place does not mean it should act on our behalf. Designing the agentic experience is, first of all, deciding how agency is divided between the human and the machine: what is delegated, what remains in people’s hands, where the line falls. This distribution is not to be suffered as a side effect of technology. It is to be designed as a choice. A system that dissolves human agency is a badly designed system, not a destiny.
III
Understanding precedes execution.
An agent does not merely process: it traverses. It moves through a situation, meets obstacles, changes course, reaches an end. It is experience in the full sense of the word — experiri, to traverse. And experience is not left to chance: it is designed. But there is no action without understanding. For this reason, before an agent can act on a system, it must be able to understand it. Understanding precedes execution, and for understanding to occur a model of the world built to an agent’s measure is needed: an ontology. Without ontology there is no experience, only error.
IV
Be actionable, not findable.
Optimizing for search engines meant being found. Optimizing for generative AI means being cited. Optimizing for agents means something else: being actioned. Not appearing in an answer, but being understood and acted upon within a process. It is a change of interlocutor, not one more rung on the same ladder. Whoever confuses actionability with visibility has not understood the transition: they are still optimizing to be seen, in a world where what counts is being usable.
V
To optimize is to minimize.
Improving an agentic experience does not only mean making it possible. It means making it efficient: reducing effort, friction, waste. Every needless step an agent takes is a cost, and the costs are many — strategic, economic, operational, environmental, social. Optimization that ignores the energy consumed in redundant loops, or the people excluded by systems that cannot speak to their agents, is not optimization. It is shortsighted efficiency. To optimize is to minimize negative impact as much as to maximize the positive.
VI
Agents demand a third system.
Digital presences have rested for decades on two systems: the operational one, which enables operations, and the informational one, which provides information. Agents demand a third: the cognitive system, grounded in knowledge models, which enables understanding and therefore experience. It does not expose data and does not open functions: it maps meaning. Building cognitive systems is the specific task of AXO, what sets it apart from every practice that came before it.
VII
AXO is a discipline, not a technique.
Agentic optimization lives on three inseparable planes: technical — the architecture that exposes knowledge; design — the shaping of intentions and flows; governance — the responsibility for what the system exposes and for its consequences. For this reason it is not work for technicians alone. It requires design, and it requires thought: philosophy, ethics, ontology. Whoever approaches it as a pure engineering problem will build systems that function but do not understand. AXO is a technical act in the same measure as it is an act of thought and of the attribution of meaning.
These principles define Agentic Experience Optimization as the discipline that designs, builds, and evolves cognitive systems: to enable the agentic experience, to minimize its costs and negative impacts, to maximize its efficiency and value — and so that, in a world where we are no longer the only ones who act, agency remains a choice and not a renunciation.