In a zero-person department, AI systems do not simply carry out specific tasks. AI agents take on all the work: from decisions about day-to-day operations to collaboration with human colleagues and communication with other teams. In the ZODIAC project, experts from three faculties at the University of Amsterdam (UvA) are pooling their knowledge to examine what changes within an organisation when a department is fully automated using AI agents. Based on a lifelike simulation, 2 researchers are exploring what a zero-person department actually means in practice, both in the short and the long term.

Why is research into zero-person departments so urgent?

Tom van Engers, Professor of Legal Knowledge Management at the UvA’s Amsterdam Law School and affiliated with the Informatics Institute: 'We live in a world in which large technology companies largely determine what our AI applications look like. Expectations about the benefits for organisations are high, and many companies are already working with AI applications. In most cases, the use of AI is limited to supporting individual employees, for example through chatbots such as ChatGPT. With agentic AI, it is possible to have complex tasks performed entirely autonomously. It is important to understand the impact of that. What risks arise? How should you deal with them? Of course, we want to benefit from AI’s potential, but we do not want to be caught out by unforeseen risks. After all, as an organisation you want to maintain control over the working methods and outcomes of autonomous AI.'

Corine Boon, Professor of HRM & People Analytics at the UvA’s Amsterdam Business School and Director of the Amsterdam People Analytics Centre: 'Many organisations feel under pressure to work with AI more and more. But the interactions between an AI-driven department and the people in the organisation raise complex questions. How do people communicate with a zero-person department? How does collaboration with AI agents work? What does such collaboration mean for employees’ skills, competencies and wellbeing? What are the implications for organisational accountability and performance?'

What do we currently know about the opportunities and risks of a zero-person department?

Boon: 'The ZODIAC project builds on studies in which we have already explored many of the issues involved. For example, we already know that having a human in the loop is very important to the success of zero-person departments. One of the key themes here is that, as a human, you need different skills to work with AI. An AI agent can take over repetitive tasks, which can increase efficiency, productivity and job satisfaction. People will then carry out fewer tasks themselves and instead supervise AI more. But we do not yet know how employees can develop the right skills for this. And we know even less about the consequences of such a changed skillset for people’s experience of work.

Another risk is that people are left mainly with complex and intense tasks. The long-term effects on employee wellbeing are unclear. Can you remain alert and perform at the top of your ability for 40 hours a week without becoming exhausted?

Another aspect of the development of zero-person departments is that it changes collaboration within an organisation. If AI takes over your boring tasks, your work may become more meaningful. But a large part of day-to-day work in an organisation is teamwork. How do you experience the social component of work if you mainly interact with AI agents?'

Van Engers: 'For routine tasks on which chatbots have been extensively trained, such as summarising texts, practical applications are already showing good results. The risks involved are probably manageable. The question is what happens as applications become more complex, as in the case of a zero-person department.

Another point requiring attention concerns responsibilities within an organisation. How do agents report on what they are doing, so that a zero-person department can be audited? How do you maintain control over the reliability of AI output, such as advice? How do you keep risks manageable? How do you prevent undesirable situations? How do you ensure timely intervention if something threatens to go wrong? How do you make sure you comply with all legal requirements? European legislation such as the General Data Protection Regulation, the AI Act and the Data Governance Act makes compliance an ongoing concern.'

Why is interdisciplinary research such as the ZODIAC project so important?

Van Engers: 'Interdisciplinary research is indispensable if we are to understand the implications of AI. The technical and human aspects of AI are, after all, inextricably linked. Agentic AI also has economic and societal impacts. You need to look at AI from different perspectives; otherwise you miss the connections between technological developments and societal issues.'

Boon: 'The current debate about entry-level jobs is telling. Will entry-level roles disappear because of AI? If so, how will employees gain the experience they need in order to progress? Solving this puzzle requires close collaboration between a wide range of academic disciplines, such as computer science, HR, finance and law.

In addition to combining disciplines, linking theory and practice is what makes the difference. At the Amsterdam People Analytics Centre, we already work extensively with public and private organisations. This means we can test the questions we are working on directly against practice. For the ZODIAC project, we are carrying out a test case together with KPMG and the shared service centre of the Dutch Ministry of the Interior and Kingdom Relations. The simulation makes it possible to follow both a zero-person department and the people around it in a lifelike setting.'

Why is the UvA’s research into zero-person departments so valuable for practice?

Boon: 'The ZODIAC project is innovative and pioneering. From an HR perspective, for example, it is unusual that we are using a simulation to investigate various dynamics. We will follow the test situation in real time, enabling us to capture different effects in detail. Through the collaboration between three faculties and two professional partners, we are able to examine in depth the questions that are currently urgent for organisations. This will allow us to produce a more comprehensive set of recommendations that will genuinely help organisations move forward.'

Van Engers: 'We are really exploring the next phase of AI. Employees are becoming increasingly accustomed to chatbots. But there has been very little research into what a more complex use of AI means for an organisation. The research gives a vision of the near future. Where are zero-person departments a good idea, and where are they not? What safeguards are needed? What boundaries do we want to set? The research outcomes will enable organisations to actively steer the direction they want to take.'

Want to know more?

Read: UvA project ZODIAC receives KHMW Stimulus (18 March 2026)

The Impact Centre involved is: Amsterdam People Analytics Centre