Deep beneath the Atlantic, Cold War submarines hunted one another through the dark. A trace of engine noise, the vibration of a pump, the rhythm of a propeller could betray a vessel that was otherwise invisible. Analysts learned to recognize these acoustic signatures and infer what was moving through the water. I became fascinated by the idea that a complex system reveals itself through the pattern of its operation. A company, I thought, might do the same through the decisions it makes.
For the past two years, Nitin Mittal, Deloitte’s global AI leader, and I have been exploring what will distinguish companies when powerful intelligence is widely available. We began collaborating on ideas about a new industrial revolution. AI was clearly the disruptive catalyst, but there seemed to be a piece missing in the story. Deloitte was confronting this issue firsthand as AI began to challenge the economics of expertise on which its business depended. Those conversations became the starting point for our book. Our answer began to take shape around the decisions that matter, and how companies distribute human and machine judgment within them.
Many of the AI maturity models we encountered measured progress primarily through adoption. They offered less guidance on how those capabilities would create a distinctive advantage. Leaders are used to big picture frameworks. Porter’s Five Forces, the growth/share matrix, the Balanced Scorecard and the Business Model Canvas are powerful because they make complex tradeoffs easy to discuss. At the more tactical level, dashboards, spreadsheets and enterprise telemetry provide a wealth of operational detail. However between those perspectives lies a gap in understanding. We wanted a design language that could connect those perspectives and help leaders work through the choices about where human and machine judgment belonged.
The familiar idea that “humans and machines should work together” sounded sensible, but left the important choices unresolved. A person could set an objective, approve every action, handle exceptions or review results afterward. Each arrangement places human judgment somewhere different. The same problem applies to insisting on a human in the loop. A well-informed person with time to intervene can improve a decision; someone approving recommendations they barely understand may add little. We became interested in the precise distribution of judgment, including when a machine should be trusted to act and where a person could make the greatest contribution.
Decisions gave us a practical place to start. As digital labor gains the ability to assess a situation and act on it, its role extends beyond completing assigned activities. Drafting a customer reply is one task; deciding whether to make an exception, offer a refund or commit the company to a remedy exposes the judgments that shape the outcome. These are choices about authority and responsibility as much as software. They give leaders something concrete to redesign.
At Covestro, the German materials company, operators at a production line in Dormagen tested a model’s recommendations until they were confident enough to ask whether they still needed to push the button. The system was then allowed to make production decisions within defined limits. Petrobras, Brazil’s largest oil and gas company, uses a different arrangement for maintenance: an AI agent recommends a response, and a human supervisor approves it before a work order is issued. The system learns from the feedback and gets smarter with time. Both applications aim to improve demanding industrial operations. In one, people establish boundaries for autonomous action; in the other, they judge a proposed intervention before it happens.
We called this pattern the Intelligence Signature: the distribution of human and machine judgment across the decisions that matter to a company’s strategy. The challenge is to make that arrangement work across several priorities at once. Giving a system more authority might help it respond faster while creating new issues about how it protects information or handles unexpected customer requests. Some choices improve several outcomes together; others require a trade-off. We organized those priorities into six domains, which I’ll explore in later articles. For now, the important idea is that the right arrangement depends on what the business is trying to achieve.
I wanted our new framework to have a visual form that people could understand, debate and change. Leonardo da Vinci’s drawings of polyhedra were one inspiration; the interlocking geometries of the Sri Yantra and the patterns of the I Ching offered others. Their appeal was the ability to hold several relationships in view at once. I imagined cognitive engineering becoming a creative discipline, accessible to someone who understands customers or human behavior as well as someone who writes software. Aesthetics mattered because an elegant visual language could invite more people into a conversation that would otherwise remain technical.
As powerful AI becomes widely available, I believe competitive advantage will depend increasingly on how a company arranges intelligence around its strategy. The same technology can support very different organizations, depending on where people apply judgment, what machines are authorized to do and how both learn from experience. The Intelligence Signature gives that arrangement a recognizable form, much as sound reveals something about a hidden submarine. Its value still has to show up in performance. Intelligence may become easier to acquire while a distinctive, effective way of using it remains difficult to copy.
Your organization already has a way of thinking, shaped by past technology choices and inherited management habits. AI is changing it with every new responsibility you delegate to a machine. How much of that thinking have you consciously designed? That question runs through Abundant Intelligence: How Digital Labor Will Rewrite the Rules of Business, the book Nitin and I have written together. Over the coming months, I’ll share more of the ideas behind the Intelligence Signature and the companies that helped us develop it.
You can pre-order the book now, and I hope you’ll join me in exploring what your organization might become.