AI is moving from answering questions to taking action. It can resolve customer issues, manage infrastructure, approve transactions and make decisions on behalf of the business. The more autonomy it gains, the more important one question becomes: do we really know what it is doing?

Many organizations are already struggling to see where AI is being used, how its decisions are made and who is responsible when something goes wrong. As agents become embedded in business processes and IT operations, traditional monitoring and governance may no longer be enough. Trust cannot rely on the assumption that the model will behave as expected. It requires visibility, clear boundaries and the ability to intervene.

During this CIONET Executive Roundtable, prepared together with Datadog, we will discuss how to stay in control as AI systems become more autonomous. Zbigniew Kamiński from Booksy will open the conversation by sharing his practical experience in this area.

We will explore:

• What changes when AI moves from recommendations to autonomous action?
• How can organizations see what AI agents are actually doing?
• Where should the boundaries of AI autonomy be set?
• Who is responsible for decisions made across models, agents and systems?
• What signals help distinguish a harmless anomaly from a real business risk?
• When can AI operate on its own, and when must a human remain in control?