AI is forcing companies to answer a question they have avoided for years

There is something deeply fascinating happening with AI adoption inside companies right now.

Technological revolutions tend to follow relatively predictable patterns. We saw it with the internet, with cloud infrastructure, with mobile, and later with data platforms: First comes experimentation, then enthusiasm, then aggressive deployment and only later does the difficult part begin: learning how to operate responsibly and efficiently at scale.

What makes AI different is the speed.

Organizations are deploying AI systems into production environments before they fully understand how those systems change the economics, workflows, and dynamics of the company itself. And in many cases, they are measuring success through the easiest signals available: number of licenses deployed, prompt volume, token consumption, or visible usage across teams.

The assumption seems to be that if AI usage grows, productivity must also be growing. But those are not the same thing.

Sooner or later, every CEO, CFO, and board reaches the same moment. The same uncomfortable conversation emerges behind closed doors:

Are we actually becoming more productive, or are we simply becoming more active?

That distinction matters much more than people realize.

AI spend becomes a problem when organizations lose visibility

I do not believe AI spend becomes problematic because the bill itself gets large. Large technology investments are normal during periods of transformation. Companies spent aggressively on cloud infrastructure long before optimization models existed. The same happened with data warehousing, SaaS adoption, and cybersecurity.

The real problem starts when organizations lose the ability to connect investment with measurable organizational improvement.

That is the dangerous territory many companies are entering now.

AI is scaling across enterprises faster than the operational discipline required to manage it. Most organizations still do not know how to evaluate AI contribution correctly. Most employees have not been trained to work effectively with these systems. And many leadership teams are still relying on management frameworks designed for a world before AI existed. Actually, Dave wrote a very interesting article about it you can check here: The AI nobody has explained us how to use

This creates a very deceptive environment because activity suddenly explodes everywhere: code, documents, tickets, protoypes,… More output appears across the system.

At first glance, the organization looks dramatically more productive.

But activity has never been the same thing as impact. AI simply amplifies the gap between both concepts.

The irony is that AI is not exposing a new measurement problem. It is exposing an old one that companies managed to ignore for years.

I am skeptical of the “10x productivity” narrative

Not because AI is not transformative. I believe it is one of the most important technological shifts we will experience in our lifetime.

What I find problematic is the way productivity is being discussed.

Most conversations reduce productivity to individual output, lines of code, tasks… But organizations do not operate as isolated individuals. On the contrary, they operate as interconnected systems.

Software development, for example, was never constrained purely by the speed at which humans typed code. The real constraints have always been coordination, prioritization, architecture, testing, communication, review processes, technical debt, organizational alignment, and decision making.

AI can accelerate one layer of the system enormously, but accelerating one layer often exposes bottlenecks somewhere else:

• You generate code faster, and suddenly review becomes the bottleneck.

• You automate implementation, and now testing slows everything down.

• You accelerate delivery, but strategic prioritization remains unchanged.

• You increase output, while organizational complexity continues to delay decisions.

This is why I do not think the winners of the AI era will simply be the companies with the highest AI adoption rates. They will be the companies capable of understanding how AI reshapes the operating model of the organization itself.

The real challenge is organizational, not technological

The most interesting shift happening right now is managerial, not technical.

For decades, companies evaluated knowledge work through proxies because actual contribution was difficult to observe directly. Time spent working became a proxy for value. Activity became a proxy for impact. Process compliance became a proxy for effectiveness.

AI starts to break many of those assumptions.

When generating output becomes exponentially easier, the scarce resource is no longer production itself. The scarce resource becomes judgment: the ability to prioritize correctly, reduce friction, coordinate effectively, and identify what actually creates value for the organization.

This is why I believe companies are entering a period where understanding organizational performance becomes strategically critical.

Not in the superficial sense of dashboards or vanity metrics, but in the deeper sense of understanding how work flows through the company, where bottlenecks emerge, whether AI is genuinely improving delivery, what quality tradeoffs are appearing underneath acceleration, and whether rising AI costs are translating into meaningful business outcomes.

Because eventually, every organization will have access to the same models.

The real differentiator will not be AI adoption itself, but the ability to operate, measure, and manage AI-enabled organizations more intelligently than everyone else.