EGGS Intelligence Report

Why larger organizations have the toughest AI problem.

There is a paradox in the adoption of artificial intelligence in large Spanish organizations: those that have invested the most in digital transformation in the last ten years are often the ones that have the most difficulty integrating AI in a way that generates real value.

It is not an accidental paradox. It has a structural cause.

The complexity that already existed

86% of managers surveyed in a recent global study by Harvard Business Review say that the organizational complexity of their companies has grown to the point of slowing growth. 70% of digital transformations do not achieve their objectives — a fact confirmed by BCG across multiple research cycles.

These numbers are not from 2020. They are from 2024 and 2025. Years in which organizations have been digitally transforming for a decade.

What the data suggests is that digital transformation, in many cases, has not reduced organizational complexity — it has amplified it. It has added layers of systems, processes, specialized equipment, and dependencies between areas. And when that complex organization now tries to integrate AI, it is not adding technology on a clean slate. You are adding it on top of an infrastructure that already has its own frictions, its own misaligned incentives, its own resistance to change.

AI does not simplify complex organizations. He exposes them.

The adoption gap

The data on actual adoption of AI in the workplace is revealing. In the Nordic countries — the most digitized in Europe — only 19% of workers use generative AI on a regular basis. In Spain, the context is comparable or more conservative.

And yet, 82% of European managers say they have deployed generative AI or plan to do so this year.

That distance between what is deployed and what is used is the real problem. It's not a technology problem — it's an adoption problem. And adoption does not fail for lack of tools: it fails because the tools have not been integrated into people's real workflows, because the change has not been managed with sufficient attention to the human factor, or because organizational incentives do not align teams with the use of the new technology.

An AI system that no one uses is more expensive than no AI system. Pay for the technological investment without receiving the return. And it creates skepticism in the organization about the real value of AI — a skepticism that is then very difficult to reverse.

What organizations that do it well have in common

In the projects we have been working on for years — in banking, in energy, in the public sector, in industry — we have identified some patterns in organizations that manage to integrate AI in a way that generates real value and sustained adoption.

_They start with the problem, not the technology. _Organizations that successfully integrate AI don't start from "we want to use AI" — they start from "we have this specific problem in this specific process, and we want to understand if AI can help solve it better than what we have now." That difference in starting point changes everything.

They design the experience of the people who are going to use it. An AI model that improves the efficiency of a process only generates value if the people who work with that process use it well. That requires research with the real users of the system — not just the decision makers who approve it — and an experience design that makes the change understandable and manageable.

_They treat responsible AI as a design requirement, not a legal notice. _Organizations operating in regulated sectors – banking, insurance, public sector – have additional pressure: the European AI Act establishes concrete obligations of transparency, explainability and human control for high-risk AI systems. Those that treat this framework as an obstacle manage it poorly and arrive late. Those that integrate it into the design from the beginning have systems that their users understand and their regulators accept.

_They measure adoption, not just deployment. _The KPI that matters is not how many AI systems have been launched — it's how many people are actively using them and what impact they have on business results. That measurement requires time and rigor, but it is the only one that distinguishes real transformation from the appearance of transformation.

The role of design in all this

There is a phrase from an in-depth interview with a company executive that we cited in our internal research process. Speaking about his design teams, he said: "Every time designers mention Figma, they are degrading themselves."

The phrase is harsh. And it has some truth.

Design in the context of integrating AI into complex organizations cannot just be a process of creating interfaces. It has to be a practice of understanding the organizational problem, aligning incentives, and managing human change. If the designer arrives when the solution has already been decided and only has to give it shape, he is late.

The AI ​​projects that work are the ones that include design — and research with real users — in the phase when the decisions that matter are still being made: what problem to solve, with what approach, with what adoption logic.

That is what we mean by strategic design in the context of artificial intelligence. And it is the most difficult — and most necessary — work ahead of us.