Why AI in companies still doesn’t work: the problem is not the technology

In recent years, artificial intelligence has become a constant presence in corporate strategies. It is often described as a driver of productivity and competitiveness, yet in practice the value generated frequently falls short of expectations. This gap between promise and results is not due to technological limitations. More often, it stems from how AI is introduced within organizations.

One of the most common misconceptions is the belief that AI adoption can happen quickly, almost by substitution: a tool is implemented and processes automatically improve. In reality, this is rarely the case.

From our experience at CHORA, it became immediately clear that AI does not enter an organization overnight. It requires time, understanding, and a gradual path. Assuming that simply adopting a tool will automatically improve a process means underestimating organizational complexity—especially in technical and engineering environments.

Understanding what AI can do before using it

Before discussing operational applications, it is essential to understand what artificial intelligence can actually do and how it works. For this reason, around two years ago, CHORA chose to start with a preliminary AI training program. The goal was not to learn how to use a specific piece of software, but to build a shared understanding of application areas, limitations, and practical implications.

This phase is often underestimated, yet it is decisive. Without a shared knowledge base, AI risks remaining a “black box”: powerful, but poorly understood, and therefore difficult to adopt in a structured way.

Only after clearly defining what AI can and cannot do does it make sense to move forward.

Established processes and resistance to change

Once the initial training phase is complete, the most complex issue emerges: processes. In many companies—and especially in technical and engineering firms—processes are the result of years of experience and countless trial-and-error iterations. They are not streamlined by nature, and changing them is not easy.

Moreover, it is not always true that all business processes can be immediately redesigned using AI. Forcing integration often leads to isolated experiments or counterproductive results. This is one of the main reasons why many AI initiatives fail to move beyond the initial phase.

When technology moves faster than the organization

Another critical aspect we have observed, both internally and when working with other organizations, is the speed at which technology evolves compared to an organization’s ability to absorb it. AI follows a clear, exponential growth curve. Models improve rapidly, applications multiply, and possibilities expand.

Business processes, on the other hand, require time: validation, testing, and progressive adjustments. This gap often generates frustration and leads organizations to look for problems in data or tools, when in reality the issue is cultural and organizational. There is a widespread feeling that “more could be done,” without being able to translate that perception into concrete change.

A starting point, not a failure

Without a gradual approach, proper training, and a conscious review of processes, AI remains an external tool. It is used sporadically, often at an individual level, without truly impacting decision-making or organizational structure.

This is where the biggest misunderstanding arises: it is not AI that fails, but the way it is adopted. Recognizing these challenges does not mean rejecting artificial intelligence. On the contrary, it means creating the conditions for a more mature and effective use.

At CHORA, we have learned that facing these limitations is part of the journey. Understanding that not everything can be automated immediately, and that change requires time, is the first step toward building real value with AI.