When a Deep Research run ends, most products hand you a long piece of writing. What you actually need to send is sometimes a report, sometimes a structure, a flow board, or a picture.
The next step with a text draft is usually to paste it into the app you already use. Wording goes to a writing tool. Levels go to a map tool. A flow means placing the boxes and arrows again. A picture gets exported into an image editor. The sources, the levels, and the layout are easy to lose in that move. What the run produced is a first draft. The editing after that is the work.
You can ask the model for another version in the chat. Moving a branch, dragging an arrow, changing one region of a picture, or adjusting a vector shape and its type are not things the chat box does. Asking for another version means describing the change all over again.
What we did
A finished run almost always needs another pass, and the edits are not the same gesture. So the result goes straight into the editor that can make that gesture, and you change the original.
The document editor is for the research report. The wording and the structure of a first draft usually move. Edit it in the same piece, and each sentence can still go back to the passage.
The mind map editor is for structure. Once a topic opens up, what you have is branches and levels. Move a branch, add a subtopic, and change the level on this map. When you need to take it with you, it can go back and forth with Xmind.
The whiteboard editor is for flows and architecture. These are boxes and arrows. The layout from the run stays. After that you drag, align, and pull overlaps apart.
The image editor is for the picture after it is generated. The usual changes are the text, the subject, the background, or one region. You change the original, and what you did not touch stays.
The vector editor is for icons, illustrations, and diagrams. These go into a document, and they get enlarged. Once it is a vector, you change the shapes, the type, and the layers on the same graphic. A normal picture can be turned into a vector and edited from there.
Finished results are in the library. Open the form this run was supposed to hand over.