Often, it is limited to writing texts or summarizing information. This is efficient and sometimes outstandingly good. Nevertheless, follow-up work is often necessary to achieve a high-quality result. This is mainly because these applications have to be guided – and this requires the user to be clear about the question and have a precise picture of the desired result.

Programming with AI: a pioneer in use

AI tools are used much more frequently in the area of software development. In our development department, programmers are already successfully using AI to support each other. Tools like Copilot do impressive work, especially with rule-based tasks, which are common in programming.

One decisive advantage is that code is structured and based on clearly defined rules. Language, on the other hand, is more complex and often ambiguous. This means that the use of AI for text-based tasks still requires careful control and expertise.

The challenge: asking targeted questions

One of the biggest hurdles in using AI is asking the right questions. The motto here is:

• specific question – specific answer
• general question – unspecific answer (even if it sounds professional)

Particularly when it comes to complex tasks, such as creating sophisticated texts or building complex program structures, it is not enough to just have a vague idea. In these cases, expertise and a clear understanding of the desired result are essential.

So-called AI agents are used to improve interaction with AI tools. These support the user by structuring the process, asking more specific questions and making the results interpretable. But now there is an exciting new development: autonomous AI agents.

What are autonomous AI agents?

Unlike traditional AI agents, which still require active interaction with the user, autonomous AI agents work largely independently, as the name suggests. Microsoft, one of the leading players in this field, recently presented such a tool.

A practical example from McKinsey:

An autonomous AI agent can be used in the recruiting process, for example. McKinsey receives a large number of applications every day that need to be processed individually. The autonomous agent analyzes the application texts, identifies open questions and uses this information to create specific answers. It then automatically generates a suitable e-mail and sends it.

This level of automation is truly new and offers enormous potential. It will be particularly exciting when such agents are connected to business objects, for example from our iCore. This enables the technical processing of data on a whole new level.

Conclusion: the future with autonomous AI agents

Autonomous AI agents mark an important next step in the development of artificial intelligence. While traditional AI tools still require input and control from the user, autonomous agents can carry out processes independently.

This not only opens up efficiency potential, but also makes AI more accessible for complex business applications. However, a technical basis such as a semantic data model is essential to make optimal use of this data. But more on that in a future blog post.

The development shows: we are only at the beginning of a new era, and autonomous AI agents could soon be indispensable tools in the working world.