Attachments

By using attachments, you can easily create documentation drafts from existing material or provide additional context to the AI model.

There is a large number of file formats that can be attached to the conversation, including images (in png, jpeg, svg format), XML, Markdown, text, Word (.docx), Powerpoint (.pptx), Excel (.xlsx), and PDF files. For Word and PDF files, any included images are also sent to the AI engine.

Images that have the width or height set larger than 2000 pixels are automatically scaled down before they are sent to the AI model, so that neither side exceeds 2000 pixels. The aspect ratio is preserved and the attached file itself is not modified, only the copy that is sent to the model is resized. This keeps large screenshots within the limits of the vision models and reduces the number of tokens they consume.

You can drag (or copy) files from your system file browser or screen captures and drop (or paste) them in the chat input area, or you can use the dedicated Attach action in the top-left corner of the chat input box. The attached files are displayed as chips at the top of the chat input area.

Clicking a chip opens the referenced document in its associated application. You can also reference an attached file in a specific part of the edited message by typing @ and then selecting an existing attached file or an already open document. Otherwise, all attachments are automatically expanded and uploaded at the end of the chat input message.

Chips are also used to present attachments in the chat conversation thread. If you use the Edit button that appears to the right of a sent message in the response area to modify that message, you can also use the Attach action that appears above the edited message to attach files and they are displayed as chips above the particular message. You can hover over a chip and use the Edit button to modify an already sent message and add or remove attachments from it.

Whenever you select content in the editor, a chip named selection is added at the top of the AI Positron's chat input area. It expands the selected content (including images) and it updates every time the selection is changed. You can reference the selection in a message by typing @selection.