The Intelligence-Centered Enterprise Is Taking Shape

Something is shifting inside the enterprise.

AI started as a tool. Then came copilots. Now agents are beginning to reason, make decisions, and act across workflows.

Yet most companies still operate much as they did before AI arrived. A customer signal enters one system. The data lands somewhere else, and someone builds a dashboard. A meeting gets scheduled. A decision gets made. Another team executes it. Eventually someone asks whether it worked.

AI can move in seconds. The organization around it often cannot.

That gap is what the Intelligence-Centered Enterprise, or ICE, is trying to name.

In the first two papers of the ICE series, we argued that the next stage of AI transformation is not about putting more AI into the enterprise. It is about redesigning the enterprise around intelligence.

The four verbs that matter

At the center of ICE is a deceptively simple loop: sense, reason, act, and learn.

An organization senses what is happening across customers, operations and markets. It reasons across those signals using data, AI and human judgment. It acts. Then it learns from what happened.

Companies already do all four. The problem is that they do them in different departments, on different systems and at very different speeds.

AI changes that equation. Imagine a retailer detecting a shift in customer behavior, reasoning across inventory, pricing, and supply chain data, taking an approved action, and learning from the outcome. Not over several planning cycles, but continuously.

That makes AI more than automation. It starts to look like an enterprise nervous system.

Four layers, one system

The ICE framework identifies four layers that have to work together: leadership and governance, enterprise architecture, infrastructure and platforms, and delivery discipline.

The first three will sound familiar to any chief data officer (CDO). The fourth is the one to sit with.

As AI makes intelligence cheaper and faster, the advantage may not belong to the company with the most AI. It will belong to the company that learns fastest.

Companies have spent decades building disciplines around software engineering, project management, cybersecurity, data governance and enterprise architecture.

What is the equivalent discipline for AI systems that keep changing after they go live? Who owns the agent after deployment? Who watches what it learns? Who decides when its permissions should change? Who measures whether it is still producing the intended business outcome?

Paper 2 in the series calls this the move from operating model to operating reality. That is where the harder part of enterprise AI begins.

The team is changing too

ICE also proposes an organizational building block: the pod.

Think small, persistent teams built around an outcome rather than a function.

A product lead sets intent and priorities. A systems architect designs and tunes the environment. An agent swarm handles parts of the sensing, analysis and execution.

The names matter less than the shift behind them. AI does more of the processing and execution. Humans move toward intent, judgment, exception handling and accountability.

AI can move in seconds. The organization around it often cannot.

There is evidence of what that might do to organizations. A study of 5,179 customer support agents by Erik Brynjolfsson, Danielle Li and Lindsey Raymond found generative AI raised productivity 14% on average. Novice and low-skilled workers improved 34%. The most experienced barely moved.

The results suggest AI does something other than make experienced workers faster. It could compress the experience curve.

Knowledge that once took years to acquire can increasingly appear at the point of work. The advantage then moves from who knows the most to which organization can capture what it knows, distribute it and keep improving it.

A different dashboard for the CDO

If the organization is becoming an intelligence system, it needs different measures. ICE proposes four.

• Signal-to-decision latency: How long does it take an important signal to become a decision?
• Experimentation yield: How many experiments become repeatable improvements?
• Cognitive leverage: How effectively are machine decisions combined with human intervention?
• Knowledge refresh rate: How quickly are models, prompts, rules and organizational knowledge updated?

They measure something traditional technology dashboards rarely capture: how quickly the enterprise is learning. That matters more when AI systems are not static.

Models drift. Data changes. Regulations move. Customer behavior shifts. Agents gain new tools.

An AI system can be running perfectly while the knowledge behind it goes stale.

From AI-enabled to intelligence-centered

Most companies today are AI-enabled. They have copilots, models, assistants and perhaps a few agents sitting inside processes designed long before generative AI existed.

ICE points toward something different. In an intelligence-centered enterprise, data, AI, people, processes and governance form a continuous loop. Intelligence does not stop at the dashboard. It moves into decisions, actions and learning.

That has implications for the CDO. For years, the job was to get the right data to the right person at the right time. AI adds another possibility: getting the right intelligence into the right decision at the right time, whether that decision is made by a human, an agent, or both.

So the next enterprise AI question is not which model to deploy or how many agents to build. It is simpler than that. How fast can your organization sense, reason, act and learn?

As AI makes intelligence cheaper and faster, the advantage may not belong to the company with the most AI. It will belong to the company that learns fastest.

The views and opinions expressed in this article are those of the author and do not necessarily reflect those of CDOTrends. Image credit: iStockphoto/peterschreiber.media