Google Drive's Android document scanner can now capture a stack of pages in a single pass, holding the camera over the pile while on-device AI separates each page, skips ones it has already got, and swaps blurry frames for sharp ones. The catch is a hardware floor: your phone needs at least 8GB of RAM.
What the update does
Google announced the change on its Workspace Updates blog in June 2026, with the rollout reported from late May. Instead of the old rhythm of position, tap, flip, position, tap, flip, you hold the camera over the stack and work through it while the app captures.
The pieces that make that work:
- Page detection and separation. The scanner identifies and isolates distinct pages as they enter the frame, rather than treating whatever is under the camera as one image.
- Auto-Best Frame, which automatically replaces a blurry capture with the highest-quality frame available.
- Duplicate detection, so a page you already caught does not appear twice.
- A pause button, for when the automation is moving faster than you are, plus the option to pull existing photos in from your gallery.
Availability is unusually broad. It went to both Rapid Release and Scheduled Release domains, and it applies to Workspace customers, Workspace Individual subscribers and personal Google accounts alike. There is no admin control, which is worth knowing if you work somewhere that likes to decide which capture tools exist.
The on-device part is the actual news
Scanning and image processing run locally. The feature works offline, and you only need a connection when it comes time to save, because saving means uploading to Drive.
That is a meaningful change and it deserves more attention than the speed. Document capture has spent a decade drifting toward "send the image to a server and get a nicer image back", which is a perfectly good engineering decision and a slightly uncomfortable one when the document is a payslip or a medical letter. Running the vision work on the handset means the intermediate frames, including the blurry ones the app discards, never leave the device.
It does not mean the finished scan stays private. It goes to Drive, that is the entire point of the feature. Local processing narrows the exposure, it does not remove it.
About that 8GB
Phones with less than 8GB of RAM fall back to the old single-page scanner. Google's stated reasoning is that real-time page tracking, blur detection and duplicate comparison need the memory, which is plainly true.
It still produces an outcome worth naming. The organisations that hand out the cheapest available handsets are frequently the ones whose staff deal with the most paper: clinic reception desks, community services, site offices, anywhere a form arrives on a clipboard. The feature that saves the most time on a forty-page stack is not available on the devices most likely to be pointed at one.
There is no villain in that story, just a hardware requirement doing what hardware requirements do. But if you are specifying a fleet of work phones this year, "8GB of RAM" has quietly become a document-workflow decision as much as a performance one.
What a faster scanner does to your files
Removing friction from capture means more capture. That is the point, and it is also how people end up with a 60MB PDF of a lease that nobody can search.
Three habits keep the speed from turning into a mess.
Run OCR on anything you might need to find again. A scan is a photograph of text, which means it is invisible to search and silent to a screen reader until the text layer exists. Our guide on making a PDF searchable covers what that involves.
Do not scan everything at maximum quality. Higher resolution is not free, and past a point it buys you nothing but file size. We went through the trade-offs in scanning resolution and DPI, and if a batch has already got away from you, compressing on Android is a reasonable rescue.
Name it at capture time. The scan you file as "Scanned Document 47" is a scan you will find by opening seventeen files in a row. Ten seconds of typing now, or four minutes of guessing in November.
One more small thing, learned the hard way by anyone who has fed a duplex batch through fast: check the page order before you trust it. Automated capture makes ordering mistakes faster too, and sorting out page order afterwards is dull work. Our broader guide to scanning with your phone covers the technique side, most of which is unchanged. You are just doing it a lot quicker now.
