Skip to main content

Message Menu

A message's icons appear when you hover over it or move the keyboard focus into it. The latest answer's menu always stays in view, dimmed until you hover over it.

User messages​

User message menu, in a row below the message, includes buttons:

  • Save: Saves the message to the Promptbook. It opens the Promptbook's New prompt form with the message text filled in, so you can name the prompt and click Save prompt
  • Edit: Allows editing the message and sending it again with Send, or leaving it unchanged with Cancel. The edited message is sent with the model and attachments selected at that moment, and a response still being written stops first. This starts a new branch of the conversation: the previous version of the message and the messages that followed it are kept, and the < n/m > switcher on the message switches back to them. The switcher can't be used while a response is being written

Assistant messages​

Assistant message menu provides several options for handling message content:

  • Copy: Copies rich-text formatted message contents to clipboard. The rich-text version is marked as AI-generated; the plain-text version is unchanged. The button shows Copied for a moment

  • Read aloud (AI voice): Reads the message aloud in an AI-generated voice, can be stopped by clicking again (Stop reading (AI voice)). The AI voice is tried first. When it is unavailable, a notice says so and the browser's basic voice reads the rest of the message:

    • "AI voice usage limit reached — available again in ...", with the wait when it is known, or "AI voice usage limit reached — try again later.", each followed by "Using the browser's basic voice."
    • "AI voice unavailable — monthly quota reached. Using the browser's basic voice.", with an Upgrade button that opens Options → Account
    • "AI voice unavailable. Using the browser's basic voice." for any other problem

    Unless you have set your own OpenAI API key, the AI voice uses your BrowseWiz plan's credits

  • Good response / Bad response: Rate the answer with a thumbs-up or thumbs-down, see Rate a response

To get another answer, edit the user message above it and send it again. The previous answer stays in the earlier branch.

In older chats, an answer that has several versions shows a < n/m > switcher to view them. The conversation always continues from the newest version.

Rate a response​

The thumbs-up (Good response) and thumbs-down (Bad response) buttons appear only on answers served by BrowseWiz. Answers from your own OpenAI key or a custom endpoint have no thumbs, and no message shows them while a response is being generated.

The first time you click a thumb, a Help improve BrowseWiz dialog asks how to share your ratings:

  • Share conversations: each rating also shares the text of the conversation up to that response, and any comment you add. Results of the Google Sheets and Google Calendar tools, files and images are left out, and page addresses are shared without their query string
  • Ratings only: only the rating is sent

Closing the dialog without choosing cancels the click, and the dialog opens again on your next click. After you choose, a click on a thumb sends the rating right away.

  • Comment: with Share conversations on, a thumbs-down opens a comment box (What went wrong? (optional)). Click Send to add your comment or Skip to close the box.
  • Change or clear a rating: click the other thumb to change your rating, or click the highlighted thumb again to clear it.
  • Change your choice later: in Options → Settings → Privacy, the Share conversations when I rate responses switch turns sharing on or off. Turning it off stops future sharing and deletes the conversations and comments you've already shared. If that can't be saved, the switch turns back on with a message under it; try again.

See the Privacy Policy for how shared ratings and conversations are handled.

Note messages​

Note message is created by dragging text onto the chat messages, above the message input. It is useful to quickly add text snippets for AI model to work on.

Note message menu, in a row below the note, includes buttons:

  • Edit: Opens the note's text in an editor, with Confirm and Cancel buttons. Confirm saves your changes to the note; the rest of the chat stays as it is and no new answer is generated. Cancel leaves the note unchanged.
  • Delete: Removes only the note; the rest of the chat stays

Tool messages​

When the AI uses a tool, the chat shows the tool's message. Its Copy result button copies the tool's result to the clipboard and shows Copied for a moment.

How answers look​

  • Markdown: answers are formatted: headings, lists, tables, bold and italic text, and code.
  • Code blocks have a copy button (Copy code to clipboard), which shows Copied! once the code is copied.
  • Math is typeset from \(…\) (inline) and from \[…\] and $$…$$ (display). A single $ is plain text, so prices such as $5 and $10 show as written.
  • Links open in a new tab. Only http, https and mailto links work; other links do nothing.
  • Images in an answer never load: each one shows as a link to the image's address instead.
Last updated on