Bookmarker - Your library. On your Mac. Sorted by on-device AI.
by•
A personal library that lives on your Mac. Save links, images, and PDFs; on-device AI sorts them. Private by default. One-time payment. Yours forever.
Replies
Best
Maker
📌
Friends, I've been working on Bookmarker since last year. It started as a web app, but I'm obsessed with speed, and the web wasn't enough. So I sunset the web app and rebuilt it as a local, blazing-fast macOS application.
▲ Local-first
Private. Stays that way.
Your whole library lives in a local database on your Mac. No cloud, no account, no sync you didn't ask for. Open any save in a blink, even on a plane.
▲ On-device AI
Smart, without the cloud.
Apple Foundation Models tag and summarize every save, right on your machine. Want deeper image vision? Add a Gemini key and pay the provider at cost.
▲ Links · Images · PDFs
One library for all of it.
A link, a screenshot, a folder of PDFs. Read in a clean view, pull palettes from any image, highlight inside any document. All in one place.
▲ Bring everything with you.
Export from the tools you're renting today and import it all into Bookmarker in a few clicks. Your bookmarks, your images, your PDFs, now living on your machine. No more paying every month just to reach your own stuff.
▲ Because your library should be yours.
No subscription. One payment, and it's yours forever.
If you've read this far: PH30 gets you 30% off through Sunday.
Thank you.
Report
saved a bunch of articles and let the auto-sort do its thing overnight, woke up to my stuff actually organized by topic without me lifting a finger. love that nothing leaves my laptop.
How does the on-device AI actually handle sorting over time when your library grows into the thousands? Curious whether the tagging stays accurate or if it ends up needing manual cleanup.
Report
Maker
@sebahattinsjly I've a library including thousands of bookmarks. Sorting stays fast as it's a local SQLite query, so thousands of items sort instantly with no network round-trip. Advantages of being offline and local-first, there is no network latency. On tagging: Apple's on-device foundation model auto-tags new links, and it holds up well. Stress tested many times; I've run imports of 5K+ bookmarks and it stays fast, accurate and reliable. ✨
Report
How well does the on-device AI handle large libraries, and does it work fully offline or does it still need to ping any servers for sorting?
Report
Maker
@tuncay3nr6 On-device Foundation models surprisingly work great, and scale well. I've got a large library with thousands of records, imported, analyzed and sorted, and it stays fast and smooth. As I've mentioned before, the advantage of being offline and local-first is there's no network latency. Your whole library lives in a local SQLite database, so sorting and browsing never touch a serve. It's a local query, so thousands of items sort instantly with zero network round-trip.
Sorting is 100% offline, always. The AI runs on Apple's on-device Foundation models, so tagging and summaries happen locally too; no cloud, no data leaving your Mac. New links get auto-tagged as you add them.
Exceptions to "fully offline": fetching a page's metadata/article when you first save a link needs the network (it's scraping the live page), and if you're on a Mac without Apple Intelligence, the AI falls back to a bring-your-own cloud key. But the core; your library, sorting, reading, search, all works offline.
One more exception: image analysis needs a Gemini key and network to generate titles, descriptions, usage ideas, color palettes and tags.
But Apple's Foundation models will keep improving, and eventually we'll have offline image analysis too.
Thanks Tuncay.
Report
Finally a bookmark manager that doesn't try to upsell me into a subscription. The on-device AI sorting actually works well, it grouped random links and screenshots into surprisingly accurate folders without me lifting a finger.
Report
Does the on-device AI still keep working properly if I save a few thousand bookmarks over time, or does it start to lag?
Replies
saved a bunch of articles and let the auto-sort do its thing overnight, woke up to my stuff actually organized by topic without me lifting a finger. love that nothing leaves my laptop.
@sercanpcke Thanks Sercan.
How does the on-device AI actually handle sorting over time when your library grows into the thousands? Curious whether the tagging stays accurate or if it ends up needing manual cleanup.
@sebahattinsjly I've a library including thousands of bookmarks. Sorting stays fast as it's a local SQLite query, so thousands of items sort instantly with no network round-trip. Advantages of being offline and local-first, there is no network latency. On tagging: Apple's on-device foundation model auto-tags new links, and it holds up well. Stress tested many times; I've run imports of 5K+ bookmarks and it stays fast, accurate and reliable. ✨
How well does the on-device AI handle large libraries, and does it work fully offline or does it still need to ping any servers for sorting?
@tuncay3nr6 On-device Foundation models surprisingly work great, and scale well. I've got a large library with thousands of records, imported, analyzed and sorted, and it stays fast and smooth. As I've mentioned before, the advantage of being offline and local-first is there's no network latency. Your whole library lives in a local SQLite database, so sorting and browsing never touch a serve. It's a local query, so thousands of items sort instantly with zero network round-trip.
Sorting is 100% offline, always. The AI runs on Apple's on-device Foundation models, so tagging and summaries happen locally too; no cloud, no data leaving your Mac. New links get auto-tagged as you add them.
Exceptions to "fully offline": fetching a page's metadata/article when you first save a link needs the network (it's scraping the live page), and if you're on a Mac without Apple Intelligence, the AI falls back to a bring-your-own cloud key. But the core; your library, sorting, reading, search, all works offline.
One more exception: image analysis needs a Gemini key and network to generate titles, descriptions, usage ideas, color palettes and tags.
But Apple's Foundation models will keep improving, and eventually we'll have offline image analysis too.
Thanks Tuncay.
Finally a bookmark manager that doesn't try to upsell me into a subscription. The on-device AI sorting actually works well, it grouped random links and screenshots into surprisingly accurate folders without me lifting a finger.
Does the on-device AI still keep working properly if I save a few thousand bookmarks over time, or does it start to lag?