We are entering a decade where we will have to learn to manage a new type of resource: agents and the unprecedented collaboration dynamic that they establish between your teams and your AI agents.
So far, the main question has been: “Which AI use cases should we launch?” “. Useful, but already outdated. The real question becomes: “Are we capable of measuring, comparing and improving the performance of our human teams and our agents with the same rigor?” “. As long as this question remains unanswered, agentic AI will neither be able to scale nor become an engine of profitable growth.
Before we go any further, let’s put the words down.
Analytics is not about “making charts”. It is a discipline that closely resembles economics: a way of observing reality to understand what produces value, what destroys value, and how to arbitrate between several imperfect choices. Where economics focuses on markets, analytics focuses on operations. It’s the art – and science – of transforming what’s really happening in the business into informed decisions.
For years, we practiced a static form of analytics: dashboards, indicators, monthly reports.
With AI agents, we are moving towards a dynamic, concrete form of analytics. Analytics is no longer content to describe performance a posteriori; it becomes the very way to orchestrate work in real time.
The first upheaval is the notion of role.
AI agents are not employees, but they will reveal whether or not your company knows how to manage performance. An agent who responds to customers or prepares quotes is no longer a simple performer: he must also be able to make certain decisions to be autonomous. This autonomy requires defining a mandate and clear, precise and measurable success indicators: what he has the right to do alone, what he must submit for validation, what he must transmit to a human.
As long as this mandate remains unclear, the results are inconsistent. The agent is criticized for decisions he should never have made, and praised for tasks that a script could have performed. Analytics then simply measure the confusion.
The second upheaval is the context.
An agent without context is an analyst locked in a room without windows. He sees numbers, but he doesn’t know what they represent. Early feedback on agentic analytics shows that real value comes when the agent understands the business vocabulary, the structure of customer relationships, the company’s priorities, not just the data patterns.
This context is based on a simple, but demanding idea: methodical trust, built on governed data, shared definitions, identified sources. Without this governance, data alone cannot transmit to the agent the context necessary to exercise their role. It is this trust based on the context (“context-driven trust”) which allows us to accept that an agent makes proposals, can execute actions, without going through systematic human validation.
The third upheaval is supervision.
An AI agent does not need an annual interview, it needs a specific performance indicator. A continuous flow where we measure not only the volume of tasks accomplished, but also the success rate without human intervention, the quality perceived by customers, the cost per task, the AI model used, the frequency of escalations, the types of errors made.
At this point, analytics ceases to be a rearview mirror. It becomes a nervous system: it detects weak signals, alerts you to deviations, highlights blind spots. When an agent multiplies the escalations on the same process, it is not only the agent that must be questioned, it is the process itself. The agents then become seismographs of our dysfunctions.
Ultimately, the issue is no longer whether you are going to use AI agents. You will, sooner or later. The challenge is to know if your analytics is ready to welcome and manage them.
If analytics remains confined to dashboards that tell the story of the past, AI agents will only add a layer of opacity to an already fuzzy system. If, on the contrary, you agree to make it a true science of hybrid work – with its hypotheses, its models, its experiments, its corrections – then the agents will become what they should always have been: decision-making partners, and not black boxes.
Monday morning, instead of asking your teams: “Which new AI use case should we launch into?” “, another question deserves to open your meetings: “How do we ensure that our employees and our AI agents have the necessary means to make the right decisions for the company, and to clearly explain the effects?” »
