Ask AI about your own books, and see where each answer came from
Updated 2026-11-05
Athena answers questions using only the files you put in your library, and every answer cites the passage it relied on. Tap a citation and you are back on that page of the book.
What it is
Most AI chat tools answer from what they learned in training, or from text you paste in. Athena works the other way round: you import your own ebooks, papers and notes, and the AI reads those. The answer comes with citations to the source and a rough location, because an AI can be wrong and the citation is how you check it.
You choose the scope of each conversation: one item, a shelf, or your whole library. A shelf of five books on the same subject becomes something you can question as a set — useful for essays, reading groups, or comparing how several authors treat the same idea.
How to set it up
- Import the files. EPUB and PDF open directly; MOBI, AZW, AZW3, FB2, TXT, RTF, MD, DOC, DOCX, ODT and HTML are converted to EPUB after upload. See Importing ebooks and files.
- Activate AI for the items you want to ask about: cover menu → Activate AI. This never happens on its own, and activation itself does not use any credits.
- Open an AI conversation from the sidebar or the dock, pick the scope, and ask. Inside the reader you can also long-press a passage and ask about it directly.
- Tap a citation in the answer to jump back to the passage it came from.
Questions that work well
- “Summarise chapter 3 in five points.”
- “Where does the author define “opportunity cost”, and how does the definition change later?”
- “How do these three books disagree about sleep and memory?” — with a shelf as the scope.
- “Explain this paragraph in plain words.” — long-press it while reading.
Narrow questions about a single item usually get sharper answers than the same question asked across the whole library.
The AI tools in Athena at a glance
| Tool | What it does | Needs a connection | Uses credits |
|---|---|---|---|
| Ask your library | Answers questions about one item, a shelf or the whole library, with citations | Yes | Yes |
| Summarise | Condenses a chapter or a whole file | Yes | Yes |
| Translate | Translates selected text instantly | Yes | Yes |
| AI dictionary | Explains a word in the context of the sentence you are reading | Yes | Yes |
| Text recognition (OCR) | Turns a scanned PDF into text you can search, copy and ask about | Yes | Uses recognition runs |
| Activate AI | Prepares an item so it can be asked about | Yes | No |
How many credits each tool and model uses, and what each plan includes, is on the pricing page — it reflects the rules currently in force, which is why we do not repeat the numbers here.
Scanned PDFs
A scanned PDF is a stack of pictures; without a text layer the AI has nothing to read. Run text recognition (OCR) on it first, and it becomes searchable, copyable and something you can ask about. One limit to know: the recognised text is not used for read-aloud — listening works on born-digital files only.
Limits worth knowing before you rely on it
- It needs a connection. Reading, notes and listening work offline in Athena; AI answers and knowledge-base retrieval do not.
- It can be wrong. Open the citation and read the original, especially before you act on an answer.
- If it says it found nothing, the item may not be activated yet, or the answer really is not in it.
- What you submit is used only to produce that answer and is never used to train a model. Some capabilities send the necessary content to third-party AI providers bound by contract; the privacy policy has the details.
Frequently asked questions
Can Athena answer questions across several books at once?
Yes. Set the scope of a conversation to a shelf or to your whole library and the answer draws on all of those items, with a citation for each source it used.
Does activating AI for a book cost anything?
No. Activation is free and never happens automatically. Asking questions, summarising and the other AI tools use credits; the current rules are on the pricing page.
Does the AI work offline?
No. AI answers and knowledge-base retrieval need a connection. Reading, highlights, notes, local dictionaries and listening with the device speech engine all work offline.
Which file formats can I ask questions about?
Anything you can import: EPUB and PDF directly, and MOBI, AZW, AZW3, FB2, TXT, RTF, MD, DOC, DOCX, ODT and HTML after automatic conversion. Scanned PDFs need text recognition first.
Is my content used to train AI models?
No. What you submit is used only to generate that answer. Some features pass the necessary content to third-party AI providers who may not use it beyond the service, as described in the privacy policy.