How it works
Four steps, and three of them happen on your phone.
1. Add a manual
Three ways, whichever suits where you are:
- Paste a link. Most manuals live as a PDF on the maker’s site. Paste the URL and Xgest downloads it.
- Share it in. Looking at a PDF in Safari or Files? Share → Xgest.
- Pick a file. If it’s already on your phone.
Xgest reads the whole thing on your device — a 532-page owner’s manual takes about a second — and files it under the thing it belongs to. One thing you own can have several manuals: the owner’s manual and the nav supplement, the use-and-care guide and the installation sheet.
If a PDF turns out to be a scan — pictures of pages with no text behind them — Xgest tells you so instead of pretending. It can’t read those yet.
2. Ask in your own words
This is the step that makes the rest possible.
Manuals don’t speak English; they speak manual. Searching your copy of the Honda HR-V manual for “orange horseshoe light” matches zero of its 532 pages. Searching for “TPMS” matches 26.
So Xgest translates first: your question goes out and comes back as the vocabulary the manual would actually print. Only the question makes that trip — not your manual.
3. Your phone finds the pages
Now the manual gets searched, on your device, using the index built at step 1. Out of hundreds of pages it picks the handful that are genuinely about what you asked — not the ones that merely mention the words in passing.
Those few pages, about five of them, are what get sent to write the answer. Around 1% of the manual. The other 99% never leaves, and never needed to.
4. The answer, with the page
You get a plain-language answer — what it means, then what to do — built strictly from those pages. And every answer cites the printed page numbers it used.
Tap a citation and the real page opens. The actual PDF, zoomable, exactly as printed, with the page number that matches the paper copy in your glovebox.
That citation isn’t decoration. Manual content can matter — brakes, towing limits, self-clean cycles — so Xgest answers only from what your manual says, cites where it read it, and tells you plainly when your manual doesn’t cover something rather than guessing.
The photo path
Same pipeline, different front door. When you photograph a symbol, Xgest works out what it is and turns that into the same kind of search — then answers from your manual, with the page, exactly as it would for a typed question.
Identifying is all the camera does. It never writes the answer. So if it misreads your symbol, you’ll see what it thought (“Looks like: amber tire pressure warning light”) right above a cited page that will set you straight.
When there’s no signal
Manuals get read in garages and basements. So:
- Reading your manuals — always works. They’re on your phone.
- Searching them — always works. The index is on your phone too. It’s a plain word search rather than a real answer, so it finds the page when your words match the manual’s.
- Answering, and identifying photos — these need a connection.
Xgest should never be worse than the paper manual in the glovebox.