WhattaRAG

WhattaRAG answers questions about your own documents, entirely on your iPhone.

Add a document — a PDF, a Markdown or text file, a link to an article, or text you paste in — and ask about it in plain language. Answers come back with citations to the exact passages they came from, so you can check the source rather than trust the model.

Everything runs on the device
The model, the search, and the document database all live on your iPhone. There are no accounts, no sign-in, no analytics, no crash reporters, and no advertising SDKs. Your documents and conversations are never uploaded, and the app works in airplane mode.

The one exception is Add by URL: when you paste a link, the app fetches that page so it can read it. That is the app's only outbound network request, it happens only when you ask for it, and the fetched text is then stored on device like any other document.

What it's good at
• Finding a specific fact in a long document and citing where it came from
• Questions about one section, one clause, one table, one paragraph
• Reference material you return to — handbooks, papers, contracts, manuals, documentation
• Working offline, on a plane, or anywhere you would rather not send a document to a server

What it isn't
It isn't a general chatbot, and it won't answer from outside your documents. Broad questions across an entire document — "summarize everything", "which is best", "how many in total" — are the weakest case, because the app reads the most relevant passages rather than the whole text. It tells you when an answer may be incomplete rather than pretending otherwise.

Built native
100% native Swift and SwiftUI, around 4 MB. No web view, no bundled runtime, no JavaScript bridge. That's why it launches instantly and why it's small.

Now in English and Spanish.