长时运行与个人智能体的「现在到底在干嘛」体验问题

Viv @Vtrivedy10原推发布于 2026.11.01

在 X 上打开原推

“what is actually happening rn” -> general feeling I have for - long running agents - personal agents like bots/dots - swarms part of good Agent UX is how well humans can understand the outputs and current work potential fix: dedicated agent compute + harness feature visualizing work at any given time for human consumption I think something as simple as an agent iteratively updating an HTML page with a log of previous work would be a good start chat is good for sending info custom UI/HTML is good for me consuming info

这条是引用着另一条写的：x.com/emollick/status/2116321075164500438