We often imagine intelligence as the ability to answer questions, but perhaps it is both simpler and more demanding. Before intelligence can answer a question, it must first understand the world in which that question makes sense.
A child hearing the word “home” does not think of coordinates on a map. An architect imagines a building. A refugee recalls a place that may no longer exist. The same word survives, yet its meaning shifts because the world surrounding it has changed.
Meaning is never contained within things themselves. It emerges from the world in which those things exist, from the countless relationships that connect them, from the purposes they serve, and from the history they carry.
Only much later do we give a name to that world: “context.”
The Power of Context
Context is so deeply woven into human experience that we rarely notice it. We do not wake up every morning and reconstruct the meaning of the objects around us. We simply inhabit a world where a cup is already something to drink from, a door is something to open, and a conversation is already filled with assumptions that everyone present silently shares.
This intuition lies at the heart of Martin Heidegger’s notion from his book, Being in Time, of “being in the world.” That is, we are not detached observers looking at reality from the outside. We exist within it. We understand because we belong to a network of meanings that precedes every thought, every decision, and every action.
The world, before being a collection of objects, is a fabric of significance.
Reality exists independently of us; a world exists only once someone inhabits it. Reality is simply what is; a world is reality as it is experienced, interpreted, and organized into meaning.
This distinction is subtle yet fundamental. It explains why two people can inhabit the same reality while living in different worlds. It also explains why organizations, although operating on the same data, often construct different understandings of what that data means.
Many believe that intelligence begins with data. But this reverses the order of things, because data, on its own, possesses no meaning. A customer identifier, a purchase order, a laboratory result, or a sensor reading — all of these are merely symbols until they become part of a coherent world.
Context is not information surrounding data. Context is what enables data to become information.
This way of understanding context is analogous to the way several philosophers have understood consciousness. First Edmund Husserl, and later John R. Searle, argued that consciousness is always the awareness of something. Human thought is never suspended in a vacuum; it is always directed toward a purpose, an object, or a meaning.
Context and Data
Enterprise information follows the same principle. An invoice is not simply a document; it is a promise fulfilled. A contract is not just a file stored in a repository; it is the formal expression of an agreement between parties. A customer is not merely a row in a database; it is someone with whom the organization has chosen to establish a relationship.
Every piece of information points beyond itself, and its meaning depends on the role it plays within a larger whole. This is where enterprise AI encounters exactly the same challenge that philosophy has explored for centuries: How can meaning be reliably understood?
Large language models (LLMs) can recognize astonishing statistical patterns. They can generate fluent language, summarize documents, and reason across enormous volumes of information. Yet they cannot reconstruct a world that has never been described to them.
Without context, intelligence becomes eloquent but unreliable. It produces answers detached from the reality they are supposed to describe.
The problem is not computational. The problem is ontological.
The role of context appears in another classical distinction, that between intension and extension. The extension of the customer concept is the set of all customers an organization serves. Its intension, something entirely different, is the list of properties that define what being a customer actually means.
Two people may refer to exactly the same set of customers while silently disagreeing on what being a customer actually means. Different business units often agree on the extension while disagreeing on the intension. Sales, finance, and customer support may all refer to the “customer,” yet each inhabits a different conceptual world. The word remains unchanged, but the meaning quietly shifts.
Enterprise Context
This is why enterprise integration has never been merely a technical exercise. If connecting systems is relatively easy, connecting meanings – on the contrary – is profoundly more difficult.
The challenge becomes even more subtle because context itself lives at the intersection of objectivity and subjectivity.
The enterprise possesses objective data. Orders exist, products exist, and payments exist. Yet every department organizes those facts according to its own legitimate perspective. Finance inhabits its own world. Operations another. Compliance another still. None of them is wrong; each simply emphasizes a different network of relationships.
The goal is therefore not to eliminate perspectives but to compose them into a coherent representation of the world the organization inhabits.
But how can such a world be represented? Certainly not by data alone. Data records events, transactions, and states, but it does not explain how those elements relate to one another or why they matter. Representing a world requires a shared system of concepts, relationships, and meanings through which reality becomes intelligible.
This is precisely the role of semantic models. Rather than describing isolated datasets, semantic models represent the enterprise as a coherent system of concepts that transcend individual applications, databases and business domains. It does not replace the multiple contexts in which the organization operates; it makes them intelligible to one another, enabling humans and AI alike to move coherently across different perspectives without losing meaning.
Introducing Active Context
This is the perspective behind Denodo’s vision of active context. The Denodo Platform’s universal semantic layer is not simply an additional component placed over enterprise data; it is the mechanism through which the enterprise continuously represents itself. It captures not only what exists, but how concepts relate, how meanings evolve, and how different contexts coexist within the same organizational world.
Context is the living structure of meanings that enable information to be understood, trusted, and ultimately acted upon because it constitutes a living model of the world the organization inhabits.
Seen from this perspective, agentic AI has not created a new problem. It has merely exposed one that enterprises have always possessed.
Intelligent agents cannot invent business meaning, and they cannot infer organizational reality from disconnected records. Before they can reason, they need a world.
This is where active context acquires its deeper significance. Every intelligent agent acts within a context, yet no enterprise is defined by a single one. Finance, operations, compliance, manufacturing, and customer service each inhabit different conceptual worlds. The challenge is not to impose one context upon all, but to enable movement across many contexts while preserving meaning. Active context doesn’t provide a single representation of the enterprise, but a coherent constellation of interconnected contexts through which AI can reason without losing its place in the organizational world.
Perhaps this is why the future of enterprise AI will not be determined solely by more powerful models. It will be determined by richer worlds.
Humans call it experience. Organizations call it knowledge. AI will call it context. Philosophy has always called it a world.
He is passionate about philosophy, which he considers an indispensable discipline for those wishing to deal with Artificial Intelligence. He also deals with non-verbal communication, offering specialized seminars and advice (https://www.decorporisvoce.com/) and, in his spare time, is a Food Blogger (https://www.trapignatteesgommarelli.com/).