AI agents at work. But who's in control?

In publishing, AI has so far mainly been used for generating text. The next step goes much further: agentic systems research independently, process content, and control connected systems, often across multiple steps in sequence.

With this automation comes a growing question of control. How can data sovereignty be preserved? How do individual AI decisions become traceable? And what does control look like when it's matched to the actual risk, instead of blanket approval or blocking of every step?

That's exactly what mintloop founder Matthias Kraus will address on October 7 at the Frankfurt Book Fair, in Hall 4.1 at the Knowledge Hub.

What the talk covers

Using a real integration with Langdock, the talk shows how agentic AI can be woven seamlessly into existing editorial and publishing systems, creating a consistent, transparent process from research through to publication.

The focus throughout is on who retains control over automated decisions, and how that control can be implemented in practice without undoing the benefits of automation.

Who should attend

The talk is aimed at publishers and publishing professionals who want to understand how agentic AI automation can be integrated into existing processes in a practical, controlled way.