Most AI applications today are task-oriented: they help summarise conversations, analyse data, answer questions, or create images. You give a command, and the computer performs a task. With the rise of Agentic AI, artificial intelligence truly takes a giant leap forward. AI agents are 'digital colleagues' capable of performing a series of tasks or even an entire process. Instead of executing a single task based on a prompt, AI agents can work independently, learn, and make decisions within a broad framework. They operate autonomously, continuously learn, and proactively make decisions to achieve set goals.
A key feature of Agentic AI is the collaboration between agents. 'Where you currently need a separate app for each task, soon you’ll have AI agents capable of independently running an entire process,' says Edwin van Bommel. ‘They do this by collaborating with other AI agents that can execute specific assignments.'
'For example, an AI agent can handle a customer query and, if necessary, immediately engage other agents to process an order, issue a refund, or solve another problem. Many agents will soon be embedded in software packages, such as an ERP system. Additionally, there will be app stores from which you can download agents. Essentially, they are "digital employment agencies" with specialized agents or robots.’
In this way, Agentic AI opens the door to a wide range of applications. AI agents can independently work on complex tasks with long-term planning, such as developing and executing a marketing strategy. They can also collaborate with other AI systems and engage them to solve specific sub-problems.
What ultimately makes Agentic AI particularly powerful is its ability to continuously learn and adapt. ‘An AI agent will no longer need to be pre-programmed to successfully sell items online', explains Van Bommel. ‘It will look at popular products, analyse YouTube videos with sales strategies, and learn from manuals and best practices on forums. By continuously optimising processes and analysing new data, AI agents refine their methods and improve their performance. And all this without a coffee break, mini-break, or holiday. Even with little to no human intervention in the future. This truly makes this technology autonomous: the AI can improve itself without human input.’
‘But that is still the future', continues Van Bommel. ‘Agents need to be trained first, and that requires a lot of work. For now, humans still determine what happens and make the final decision. Additionally, agents record what they do and can increasingly explain the reason for a decision. This so-called “reasoning” allows you to continue monitoring and adjusting the agents.’
Nevertheless, the technology opens up a world of new possibilities for businesses and organisation. Agentic AI has the potential to fundamentally change business processes. A few concrete examples:
The benefits of Agentic AI are not limited to digital applications and services. Connecting to the physical world is also straightforward. A good example is found in the maintenance sector. Van Bommel says, 'Imagine a factory where machines and robots operate continuously. Thanks to sensors, these machines collect real-time data on vibrations, temperature, and wear. An agentic AI system analyses this data and can detect subtle deviations that indicate a potential defect. Based on this, AI can automatically generate a maintenance request, schedule an engineer, and even order the necessary parts from the supplier. This prevents unplanned downtime, lowers maintenance costs, and optimises human resources. And thus, makes the entire production process more efficient and reliable.'
Van Bommel emphasises that companies embracing this technology will gain a competitive advantage. 'Agentic AI delivers massive efficiency gains. Companies that start experimenting now will reap the benefits in a few years.'
We are experiencing a historical milestone: partly due to the rise of Agentic AI, the boundary between human and machine is fading definitively. Mustafa Suleyman, co-founder of DeepMind, recently noted this, and Van Bommel concurs: 'People engage in conversations with AI that seem so natural they don't always realise whether they are speaking with a human or a machine. This shows how far technology has advanced.'
Van Bommel stresses the importance of a responsible approach. 'The more autonomously these digital assistants operate, the more important it becomes to maintain control over their choices and underlying principles. Consider the sources they consult and potential biases in checks and samples. Ethics, transparency, and reliability are inextricably linked here.'
'Companies that properly integrate AI agents will benefit from immense efficiency gains. But without clear guidelines and responsible use, it also brings risks, even more than traditional AI. Therefore, it is crucial to start considering this strategically now. By investing in Responsible AI and formulating a clear policy, organisations can leverage the advantages while mitigating the risks.'
Despite rapid technological advancements, many companies lag in adoption. This widens the gap between technical capabilities and the extent to which companies and organisations apply them: more and more opportunities remain unexploited. 'The business sector is already lagging behind what technology can do, and that gap will only grow in the near future,' warns Van Bommel. 'To realise value with AI, people need to change their behaviour. This takes time. Companies that start with AI now also ensure that the use of AI becomes normal for their people. Hence, "early adopters" are difficult to catch up with. Thus, my advice is: start now, experiment, and dare to make mistakes. There is already so much potential with what is currently on the market. And this is just the beginning.'