Documents
Annotate for the agent
Mark up what you see and send it to the agent with a note for each mark.
Pointing is often faster than explaining. Annotate takes a picture of the open document. You mark it up, say what you want at each mark, and send it in chat.
Mark up a document
- 1
Freeze the view
Open the document. Click Annotate for the agent, the highlighter in the document header. The page freezes as a picture over the app.
- 2
Mark what you mean
Pick a tool from the toolbar and draw on the picture. Choose from five colours.
- 3
Say what to do
After each box, oval or pen line, a small box asks What should the agent do here? Type your note and press Enter. Press Esc to skip it.
- 4
Attach it
Click Attach to chat. The picture joins the chat message box. Add any words you like, then send.
The tools
| Tool | Key | Use it to |
|---|---|---|
| Arrow | A | Point at one thing |
| Text | T | Write a label on the picture |
| Rectangle | R | Box an area, with a note |
| Oval | O | Circle an area, with a note |
| Pen | P | Draw freely, with a note |
| Blur | B | Hide names and numbers |
Undo and redo sit on the toolbar. Click a number on the picture to change its note. Cancel or Esc leaves without attaching anything.
What the agent gets
Each box, oval and pen line gets a number on the picture. Your notes are not drawn on it. They go into your message, matched to those numbers.
Before you send, the notes are listed under the picture in the message box. Check them there.
Blur pixelates the area in the picture itself. The agent gets only the pixelated version.
Where it works
It works on notes, Books, code, skills and uploaded files such as images, Word and PowerPoint. It also works on dashboards the agent makes. It does not work on PDFs yet.
Open the chat panel, and pick a model that can read images. You can attach one annotated picture per message.