AI in business: key lessons from the report “Superagency in the workplace”

In January 2025, McKinsey published the report Superagency in the workplace: Empowering people to unlock AI’s full potential, a clear snapshot of how employees and leaders were approaching AI adoption. Today, nine months later, the question is unavoidable: Have we learned anything since then, have companies really changed their ways or are we still stuck in the same contradictions?

AI in the enterprise remains an uncomfortable mirror: it reflects the enthusiasm of employees, the caution of leaders and the huge gap between expectation and reality.

In this article we analyse the most compelling data from the study and reflect on what it should inspire in leaders and companies that do not want to be left behind.

“94% of employees and 99% of the C-suite say they are familiar with generative AI.”

The first reading is optimistic: AI in business is no longer a mystery. Most employees and managers know what it is and perceive its relevance.

cBut this is where the first contradiction arises. Familiarity does not mean adoption. Having AI in everyday conversation is not the same as having it in critical processes. It is like knowing the theory of digital transformation and still managing projects on Excel spreadsheets.

The reflection is clear: cultural resistance is no longer an excuse. The problem is not in convincing people that AI matters, but in enabling them to use it effectively, safely and with real impact.

“Only 1% of companies consider themselves mature in AI deployment”.

This should set alarm bells ringing at board level. After years of investment, the reality is that hardly anyone feels mature in AI.

The explanation is uncomfortable: companies have remained in pilot territory. McKinsey portrays organisations with tests, demos and case studies, but without scaling. It’s what you might call decorative AI: it looks good in presentations, but it doesn’t change outcomes.

The parallels with other areas are obvious. No company would dare to boast of a “cybersecurity pilot” when standards such as ISO/IEC 27001 require robust systems. Nor should an AI pilot be sufficient without a governance framework, clear metrics and defined accountabilities.

Maturity is not a question of tools acquired, but of how much economic and operational value is integrated on a sustained basis. And that 1% reminds us how far away we are.

“Leaders think only 4% of employees use AI for more than 30% of their work; employees say it’s 13%.”

This is where the most dangerous perception gap emerges. Managers believe that usage is marginal, while employees claim to use AI to a much greater extent.

The risk is obvious: if management thinks it is hardly used, they will not prioritise investment in training and support. In the meantime, teams will continue to use AI on their own, without security policies or quality criteria.

McKinsey confirms it with data: we are facing shadow AI, a hidden adoption comparable to what shadow IT once was. The workforce is experimenting and taking advantage of these tools, but without corporate control.

Reflection: talent comes before strategy. Ignoring this reality not only holds back innovation, it also erodes the credibility of leaders.

“48% of employees say that formal training would increase their use of AI”.

Half of the workforce is asking for something very specific: structured training. We are not talking about motivation or attitude, we are talking about clear resources.

The data is revealing because it shows where the bottleneck is. Employees want to use AI, but they need a serious learning framework. And yet more than 20% of companies admit that they offer zero formal support today.

Here is the opportunity. Just as frameworks like ITIL standardised IT service management, organisations must standardise AI training. Not with one-off lectures, but with practical programmes that connect to real business processes.

“Employees’ top concerns are cybersecurity (51%), inaccuracy (50%) and privacy (43%).”

Enthusiasm for AI does not eliminate doubts. McKinsey identifies the three big shadows: security, accuracy and privacy.

The interesting nuance is that, despite these concerns, 71% of employees trust their company to deploy AI ethically. This trust is a strategic asset. But it is fragile: one mistake is enough to turn it into mistrust.

The reflection here is that AI cannot be deployed without robust governance frameworks. Just as the GDPR transformed the way data is managed, the future European AI Act will set the rules of the game in AI. Companies that integrate security, privacy and auditing measures now will be better positioned.

“47% of C-suite leaders believe the pace of development is too slow.”

The paradox is clear. Employees use AI more than leaders realise, but leaders, in turn, feel that adoption is too slow.

Why this contradiction? McKinsey explains it with two causes: lack of talent (46%) and lack of resources (38%). In other words, the brake is not in the will, but in the capacity to absorb technology.

The lesson is hard: it’s not just about acquiring software licences. It’s about redesigning the organisation to make AI part of the processes. And that requires skilled talent and partners to help accelerate the roadmap. Without them, the perceived speed will always be insufficient.

“Sales and marketing (28%) and software engineering (25%) account for much of the economic value of AI”.

If the question is “where to start”, the McKinsey report offers a clear answer: sales, marketing and software development are the areas where AI brings the most immediate economic value.

The reflection is that it is not about embracing everything, but about focusing on early wins. Optimising campaigns, personalising customer interactions, accelerating development cycles: that’s where the tangible ROI is.

What is surprising is that many companies continue to invest in more experimental cases, while neglecting the functions that account for 50% of the value. Prioritisation is not optional; it is the difference between an interesting project and a real transformation.

Conclusion: lessons that cannot be ignored

The McKinsey report forces us to look in the mirror.

• AI in the enterprise is well known, but hardly mature.
• Employees use it more than leaders realise, generating hidden adoption.
• Formal training is the most demanded lever.
• Trust exists, but it requires strong governance.
• The economic value is identified, it just needs to be prioritised and scaled up.

The conclusion is simple: familiarity is no longer enough. The competitive advantage will lie with those who move from pilot to maturity, with leadership that combines vision, talent and ethics.

Do you want to find out how to move forward on this path? Contact us at Pasiona and let’s work together on a diagnosis and action plan adapted to your sector.