1 paper · 1 filter
Linyi Yang, Yixuan Weng
Current deep-research agents run in a ''fire-and-forget'' mode: once started, they give users no way to fix errors or add expert knowledge during execution. We present ResearStudio…