Buki identifies all the books in a picture you see online. One click on X, a right-click anywhere else, up to 20 books per photo. AI reads the cover, OpenLibrary checks it, and the book goes on a shelf in your browser. No account. Ten catches free.
I kept seeing books in twitter posts, bookmarking them, and never finding them again. Buki is my fix: click a picture, it names the book, and you file it on a shelf. How it works: the picture and the post's text go to a vision model together, because the caption often makes an unreadable cover legible. The guess is then checked against OpenLibrary rather than trusted. I tried local OCR first (Tesseract) and it got about 5% of real covers. No account and no database. The shelf lives in chrome.storage.local. Catches go through a small edge proxy, or straight to Gemini with your own key. Limits: OpenLibrary has no cover for roughly 4 in 10 books I tested, so Buki draws one. A photo of many books works but is slower. 10 free catches, then $4 a month or $29 a year. Not open source. I'd like to hear where it fails.