What is private document AI?
Private document AI reads, summarizes, and answers questions about your files without uploading them anywhere. Inscribe imports PDF, Word, Excel, PowerPoint, text, and images, then summarizes and answers questions about them on-device. The document, the analysis, and your questions all stay on your phone or Mac, with no cloud AI provider in the loop.
Most chat-with-PDF tools work the opposite way. You upload a file to a server, their cloud model reads it, and your document sits on someone else's infrastructure. For a contract, a medical result, a financial statement, or an unpublished draft, that upload is the whole problem. On-device document AI removes it.
Is chat-with-PDF private if it runs on-device?
Yes. When document AI runs on-device, the file is never sent to a server to be read. Inscribe processes the text, tables, and images locally, so summaries and answers are generated on the same machine that holds the file. There is no cloud AI or transcription provider reading your documents.
To be precise about what that claim covers: reading, summarizing, extracting, and Q&A all run 100% on-device. The app offers optional iCloud sync that is off by default and anonymous, no-account usage analytics. Your documents are not uploaded for processing, which is the part that matters when you ask an AI about a sensitive file.
What you can do with your documents
- Summarize a long PDF or report in a concise paragraph, bullets, or a structured outline.
- Ask questions in plain language, like what the termination clause says or which line item grew fastest, and get an answer with a citation to where it came from.
- Extract the key numbers, dates, names, or action items instead of scrolling for them.
- Read images with on-device OCR, so a photographed page or a screenshot becomes text you can question.
Ask across your whole library, not one file at a time
Single-file chat is useful, but real work spans documents. Ask Inscribe answers across your entire library at once: notes, transcripts, PDFs, images, and projects. You can ask a multi-turn question, follow up, and get citations back to the source, all on-device. That turns a folder of files into something you can interrogate, not just store.
Projects that file themselves
Group related files into a project and ask project-level questions, so a case, a class, or a client becomes one place you can query. Smart filing suggests where new documents belong, which keeps the library searchable instead of sprawling.
Private document AI vs cloud tools
| Capability | Inscribe | Cloud chat-with-PDF tools |
|---|---|---|
| Where the file is read | On your device | Uploaded to their servers |
| Works offline | Yes | No |
| File types | PDF, Word, Excel, PowerPoint, text, images | Usually PDF only |
| Ask across many files | Yes, whole library | Often one file per chat |
| Account required | No | Usually |
| Answers cite the source | Yes | Varies |
Which model reads your documents
You pick the on-device intelligence: Apple Intelligence, or a downloadable offline model, Qwen 3 0.6B, that runs locally after a one-time download. Auto picks the best available for your device. None of these calls a cloud model, so there is no GPT, Claude, or Gemini behind the scenes reading your files.
Who this is for
Lawyers reviewing a contract they cannot upload. Clinicians reading a result on a personal device. Analysts pulling figures from a spreadsheet. Students turning a dense reading into a study sheet. Anyone who wants to ask a document a question without first handing it to a company they do not control.
Export what you find
When you have the answer, export to PDF or text with an optional summary, the Q&A, and the full document text. Action items can go to Apple Reminders. The output leaves on your terms, not because the tool needed the file in its cloud to begin with.
Getting started
Inscribe is free to download and has a free tier, so you can import a file and try a summary and a few questions before deciding on a plan. There are no per-minute fees, and a Lifetime option means no subscription. Import a document, choose a model, and ask it something.
Related reading
Frequently Asked Questions
Is private document AI really private?
Yes. Reading, summarizing, extracting, and Q&A run 100% on-device with no cloud AI provider. Your documents are not uploaded for processing. Optional iCloud sync is off by default and analytics are anonymous.
What file types can I chat with?
PDF, Word, Excel, PowerPoint, plain text, and images. On-device OCR turns a photographed page or screenshot into text you can summarize and question, all without an internet connection.
Can I ask questions across many documents at once?
Yes. Ask Inscribe answers across your whole library, notes, transcripts, PDFs, and images, with multi-turn follow-ups and citations back to the source, all processed on-device.
Does it work without internet?
Yes. Once your chosen on-device model is downloaded, document summaries and Q&A run offline. The file never leaves your device to be read.
Get Inscribe
Private, on-device transcription with AI summaries and cross-library Q&A. Works on iPhone, iPad, and Mac.
Get Inscribe on the App Store