Point AI at a whole book. Import an ePub or PDF and MojoPad can read the entire thing on your Mac and help you make sense of it — summarize it, map its structure, pull its claims (with receipts), connect it to your own notes, or simply have a conversation with it.
Bring a book in the usual way — drag an .epub or .pdf onto MojoPad, or use Import (see Importing). Then drop the book onto a Canvas as a card (or use its file card). The card carries a ✦ menu — that's where the book's AI powers live. Reading and indexing the book happen on this Mac; if you have turned on Bring your own model and permitted this wiki, the passages behind a question you ask go to your provider. The book's text never leaves your Mac.
A long book is far bigger than any model reads at once, so MojoPad works in passes — reading the book in chunks, distilling each, then combining the results (the same map-reduce it uses for multi-note Synthesis). That's why a big book takes a moment: it's genuinely reading the whole thing. Answers and claims carry deep-links that open the reader right at the source. DRM-protected books can't be read — MojoPad needs the actual text.
When a part stops short. In a long read, one part can run out of room or stop arriving. The lines it did finish are used and the reading carries on. (The card it deals does not yet say how many parts stopped short.) A part that fails outright, or finishes nothing, ends the job and says why. An Ask the book answer that stops short is still dealt as a card, with a Not everything worked. note in it. See When an answer stops short in Local AI.