From booking complex yet perfectly priced itineraries to delivering personalized health diagnoses that vastly outperform even the most cogent physician, your AI agents have us starstruck, bedazzled, and breath-baited. Again, the reasons are myriad, ranging from the complexity of the tasks at hand to the legal challenges of employing user agents that aren’t exactly humans. Add in a top-down model of user agents determining our preferences, our purchases, our health care choices, and our basic understanding of the world? Let’s stick with my premise: Creating centralized user agents – agents that live on and are controlled by a platform like Amazon or Meta or OpenAI – is a terrible idea that will never work. Running user agents for billions of people requires an ungodly amount of fresh, structured and parameterized data.