Rahild Neuburger

Ms Neuburger, when does an AI assistant become an agent capable of acting independently?

Rahild Neuburger: An AI assistant becomes an autonomous agent when it no longer merely processes individual, clearly defined requests, but instead independently pursues a goal, plans the necessary steps and carries out the relevant actions. Whilst a traditional assistant reacts to specific inputs, summarises information or makes suggestions, an AI agent can independently determine which next steps are required, which tools and data sources to use, which sub-tasks may need to be delegated to other agents, and when a task is considered complete. This may also involve accessing corporate systems, initiating processes, verifying information and adapting its own approach to new situations.

The key factor here is the degree of autonomy. This ranges from simple assistance functions to agents that handle multi-stage tasks and processes largely independently. When several specialised agents work together in a coordinated manner, this is referred to as a multi-agent system.

However, even autonomous AI agents or multi-agent systems require clear boundaries. Access rights, decision-making powers, control mechanisms, escalation rules and the distribution of tasks between humans and AI must be clearly defined. Acting autonomously therefore does not mean complete independence, but rather the ability to act within a clearly defined organisational and technical framework.

What opportunities does the use of AI agents offer for businesses and employees?

Rahild Neuburger: AI agents can help businesses make not just individual tasks, but entire process chains more efficient. They can consolidate information from various systems, coordinate tasks, carry out analyses, prepare decisions, and independently trigger and execute defined actions. This helps to speed up processes, reduce administrative burdens and ease the workload on staff. Furthermore, AI agents can work round the clock and ensure greater speed and consistency, particularly when it comes to data-intensive or repetitive tasks.

For staff, the potential lies primarily in being relieved of time-consuming tasks and routine processes. Research, documentation, data preparation, handling standard enquiries or certain organisational tasks can be taken over by agents to some extent. The time saved as a result can be used, for example, for more complex activities where human judgement, creativity and strategic thinking are particularly important, for communication, personal development or even rest and relaxation. This becomes increasingly important, particularly when the use of AI agents leads to a growing workload.

Ultimately, AI agents can expand employees’ knowledge and skills if, through interaction and collaboration with them, employees are able to carry out more complex and challenging tasks than they could before the AI agents were introduced. For example, employees can perform complex data analyses or draw on information from vast knowledge bases without needing to be specialists in every field themselves.

In this respect, the greater benefit may not stem solely from automation, but primarily from the fact that, with the help of AI agents, employees can act more quickly, make more informed decisions and handle broader areas of responsibility. The more successful we are in developing AI agents tailored to individual needs, the greater the benefit will be.

What is needed to ensure that AI empowers staff?

Rahild Neuburger: For AI to empower staff, simply providing new AI tools is not enough. First and foremost, it is crucial that employees understand how AI works, what tasks it is suited to, and where its limitations lie. This requires targeted training, practical workshops and sufficient scope to trial new applications independently and within their specific work context.

Equally important is the ability to critically scrutinise AI-generated results, check them for plausibility and cross-check them against one’s own specialist knowledge. Employees should be explicitly encouraged to trust their own judgement – even when AI appears to be faster or, at first glance, more convincing.

Equally important is the ability to critically scrutinise AI-generated results, check them for plausibility and cross-check them against one’s own specialist knowledge. Employees should be explicitly encouraged to trust their own judgement – even when AI appears to be faster or, at first glance, more convincing.

At the same time, clear organisational guidelines are required. These include transparent rules for handling data, clear lines of responsibility and defined criteria for verifying results. AI should not be seen as a substitute for specialist knowledge, but rather as a tool that specifically enhances human capabilities: it can take on routine tasks, process information more quickly, improve access to knowledge, open up new avenues for action, support individual learning processes and help to better organise the daily workload. This aspect is of particular importance, especially when the use of AI simultaneously leads to a greater concentration of work.

Ultimately, AI has an empowering effect when employees are allowed to try out AI tools, can develop AI skills in a task- and context-specific manner, and when an open AI mindset is fostered within the organisation – one that encourages an open and critically reflective engagement with AI. In this respect, it is not merely a question of qualifications, but above all of shaping the organisational and cultural framework.

The white paper ‘Generative AI and AI agents in business: from assistant to key player’ (in German) is available to download here.