Skip to content

Strategy & Organisation

AI in organisations calls for more than technology

Colourful notes with ideas and questions on a wall.

AI is becoming part of work faster than many organisations can put arrangements in place around it. The question is not only what the technology can do, but who remains responsible for what is done with it.

A new feature appears in familiar software. An employee tries an AI tool on their own initiative. A supplier adds a recommendation to an existing process. Often, no major decision has preceded it, yet something changes in how information is used and work is prepared.

Technology can therefore already be in use before an organisation has decided how it wants to approach it. Who reviews the output? Which data may be used? Who is authorised to set a recommendation aside? And what happens when the result is wrong?

AI changes the organisation too

The conversation about AI often begins with possibilities. What can a system analyse, summarise or predict? Once those capabilities become part of everyday work, the question shifts. It is not only the technology that changes. Information flows, the preparation of decisions, roles, the allocation of tasks and responsibility can shift too.

A recommendation previously prepared by a colleague may now be developed partly by a system. A manager receives an alert sooner, but not automatically more context. An HR process seems more efficient, until someone has to explain how a result was reached. This is precisely where AI becomes an organisational issue.

Who remains responsible?

There is an important difference between a recommendation and a decision. AI can process information, identify patterns and support a proposal. That does not remove responsibility for what an organisation does with it.

Human oversight therefore means more than having someone give approval at the end. That person needs the authority to take a different course, enough information to understand the result, the competence to assess it critically and the time to intervene meaningfully. Without those conditions, what is a human decision on paper may, in practice, be little more than a confirmation.

Who in your organisation is authorised to question a system’s proposal, and what happens when they do?

Understanding where AI is being used

Not every use of AI is recognised as such. There are formal applications, but also AI features in existing software, suppliers’ tools and individual applications that employees discover themselves. What is sometimes called shadow AI does not necessarily arise from bad intentions. People are simply looking for a workable way to do their jobs.

The organisational issue can therefore begin before the conversation about governance. An organisation can only align responsibilities once it knows where AI is already being used, for what purpose and with which data. An overview is not an end in itself, but a starting point for making the conversation concrete.

AI literacy is more than learning to write prompts

Learning what a good prompt looks like helps someone use a tool more effectively. That is useful, but says little about whether they can use the result appropriately. AI literacy also means understanding capabilities and limitations, checking outputs, recognising sensitive data and knowing when human judgement is still needed.

For leaders, this raises an additional question: how do you guide the use of something you cannot always fully understand technically yourself? Employees need to know which applications are appropriate to their role and who they can turn to when in doubt. Trust grows not by approving everything, but by making it possible to discuss how AI is used.

The provisions of the European AI Act on AI literacy have applied since 2 February 2025. Providers and deployers must take measures to support that literacy. The current text does not require them to guarantee a specific individual level. Here too, the organisational task remains important: people need the space and support to learn how to work responsibly with AI.

When AI and people come together

When AI plays a role in recruitment, selection, performance measurement, task allocation, career decisions or people analytics, the outcome directly affects people and their work. Transparency, explanations and the ability to question a decision then take on particular importance.

Not every use of AI by HR is automatically high-risk. Annex III of the AI Act does, however, list certain applications for recruitment and selection, decisions on employment relationships, promotion or termination, task allocation based on individual behaviour or personal characteristics, and the monitoring or evaluation of performance and behaviour. Whether an application falls under the high-risk rules depends, among other things, on its intended purpose and actual use. That calls for both legal and organisational assessment.

Under the European timeline, the high-risk rules apply to the applications listed in Annex III from 2 December 2027, and to systems covered by Article 6, paragraph 1, and Annex I from 2 August 2028. The transparency obligations in Article 50 have applied since 2 August 2026. This article provides neither a legal classification of any specific application nor legal advice.

Scaling up AI calls for governance

As use grows, simple questions become more important. Who may use which application, and for what purpose? With which data? Under whose responsibility? What happens if there is an incident, and how are a supplier’s additions or changes assessed? The aim is not the most comprehensive rulebook possible, but clarity that people can use in their work.

Europe is seeking to combine AI innovation with the protection of fundamental rights, privacy, transparency, human oversight and responsibility. That attention need not slow adoption. Trust and governance can instead become conditions for embedding applications sustainably, because people know where they stand and where they can intervene.

Four perspectives, one organisational issue

At Almuruna, we explore these connections through four existing domains. Leadership & People looks at trust, behaviour, leadership and adoption. Strategy & Organisation looks at governance, roles, decision-making and organisational design. Academy focuses on AI literacy and the capabilities people and leaders need to use AI responsibly. Almuruna One focuses on the responsible use of data and AI in people processes.

Not every issue calls for all four perspectives. But technology need never be the whole answer to something that also involves people, shared arrangements and organisational choices. Data and AI highlight signals. People decide.

Start with a clearer view

An organisation can only provide focused direction once it understands where AI is being used, which responsibilities come with it and where the main uncertainties lie. The quick scan is a brief preliminary assessment, not a legal or technical audit.

What we draw on

What does this mean for your organisation?

An insight only becomes meaningful when it helps you look differently at what is happening in your own organisation.