In this talk, we'll look at how agentic systems are designed today and by whom. We'll explore Claude Code as our working example. We'll annotate together the main design choices and how they affect the privacy, security and user agency / sovereignty.
Then we'll shift and ask, is this how these systems must work? We'll investigate participatory design and AI as counter examples to today's systems. What could agentic systems do if they were designed by us and for us? What would be different and what might we keep?
I expect vivid discussion and imaginations, so bring wild ideas and questions.
