There's a layer of the agent conversation that rarely gets discussed: what actually happens when agents ship to large, non-technical organizations, and how the interface patterns change once real teams start routing their work through an agent every day.
Most agents people are building right now are single-player. You chat with them in an IDE, maybe in Slack, but the experience is fundamentally one person, one context window. ClickUp is built for teams working together, which means we've had to figure out what agents actually look like in a multiplayer context with everyone tagging the same agent, workflows being assembled and handed off across people, and the task board slowly becoming something nobody opens anymore.
In this talk, I'll share what we've learned building for that environment at scale, the UX patterns that emerged, the memory and context approaches that held up (and the ones that didn't). I'll also cover why generative UI has become central to how we think about agent output when the audience is a whole non-technical team rather than a single developer.
