Engineering truly autonomous AI agents

Most AI agents still wait for us. We give them a task, they reason, call a tool, return a result — and stop. But what happens when we remove the human from deciding what happens next?

In this session, we will go beyond tool-calling agents and explore what it takes to engineer systems that can operate autonomously: pursuing goals, observing their environment, reasoning about what is happening, planning their next actions, delegating work to other agents, acting through tools, and continuously deciding what to do next.

Through a live multi-agent system connected to a content platform, we will see how content can become more than data an agent reads and writes. Changes in the platform can become signals that agents observe, reason about, and act upon — triggering collaboration between specialized agents without waiting for another human prompt.

Along the way, we will explore the engineering challenges that emerge as we increase autonomy: coordination, state, memory, permissions, guardrails, human approval, and knowing when an autonomous system should stop.

The goal isn't to build agents that can do everything on their own. It's to understand how to engineer systems that can decide what happens next — and where we still want humans to make that decision.

Delivery Options
- Duration: 30–60 minutes (flexible)
- Language: Danish or English
- Level: Inspiration (300)