Reader AI Sidebar

The AI sidebar opens on the right of the reader (click the chat button in the reader; click it again or the X to close). Use it to ask questions about the PDF you are reading.

Asking questions

Type a question, or select text in the PDF and choose Ask AI - the selected passage is added to your question automatically.

Context you control

By default the AI uses the text of the PDF as its context. Once you have added annotations to this paper, two extra toggles appear so you can include them too:

• PDF Content - the text of the paper (used by default).
• Highlights - your highlights in this PDF (appears once you have some).
• Notes - comments attached to your highlights (appears once you have some).

Pick which chat model answers from the model menu.

MAX (full context)

By default, when embeddings are enabled, Sciwand sends only the passages most relevant to your question. The MAX toggle instead sends the entire PDF, for deeper, whole-paper answers.

Because it sends the full text, MAX uses a lot more tokens - for a typical research article, often around four times as many as the relevant-passages mode - so it costs more and runs a little slower. Turn it on when you need the whole picture, and leave it off for quick questions.

Ask about a figure or region

Use the Snip button to drag a box over a figure, table, or any region of the page. The selection is attached as an image to your question (a vision-capable model is used).

Citations

Answers include citations that point back into the paper - click one to scroll to and highlight that sentence in the PDF.