FetchMark

FetchMark is a bookmark manager that keeps a real copy of whatever you saved. Instead of storing a link and hoping the page is still there later, it archives the article itself, strips out the ads, navigation, and trackers, and files a clean readable version in a library you control. The problem it goes after is link rot, where a bookmark folder built over years quietly turns into a list of dead URLs, pages that went behind a paywall, and posts that got edited into something else. A saved copy sidesteps all of that, because what you read later is what was actually published when you found it. It's aimed at the gap between a bookmark, which is really just a promise to come back, and a proper archive, which is what you need once the source is part of your work.

Saving works two ways. You can paste a URL into the composer, or use the Chrome extension to capture a page in one click without leaving the tab. Either way the archive is permanent, and the article shows up in a reader view built around typography rather than the original page's layout, which is the practical difference between a saved link and something you'll actually finish reading. Everything in the library is searchable full text, so a half-remembered phrase from six months ago is usually enough to find the piece it came from. The composer also means anything can be saved from a phone browser or a machine without the extension installed, since a URL and a paste is the whole requirement.

The reading tools go past storage. Highlights pull quotes out of an article and notes attach your own thinking to them, so the library turns into a research file instead of a pile. AI summaries are one click and produce structured takeaways, which is most useful as triage, deciding what deserves a full read and what can be skimmed and set aside. For a backlog that grew faster than the reading did, that's the feature that makes the whole stack approachable again rather than a source of guilt. Summaries come out structured rather than as one block of prose, which makes them scannable when the point is to sort twenty saved pieces into read now and read never.

Getting data in and out is deliberately easy. The universal importer pulls libraries across from Pocket, Instapaper, Raindrop, or a plain CSV, which matters given how many people had to find a new home after Pocket wound down. Going the other way, articles export as Markdown files that drop straight into Obsidian, VS Code, or anything else that reads plain text. Saved articles are also available offline, so a stored library still works on a plane or a bad hotel connection, which is exactly when a read-later app is supposed to earn its place. That combination, an importer on one side and Markdown on the other, is what makes trying it cheap, because a library that came in easily can leave the same way.

Privacy is handled at the database level, with libraries kept private through row-level security and export available at any time, which the company frames as an anti-lock-in stance. The audience is readers who treat saved articles as working material rather than trivia, so researchers, students, writers, and anyone keeping a running file on a subject. It's less suited to someone who just wants a tidier bookmarks bar, since the archiving, search, and summaries only pay off when there's a real reason to come back to something later. Row-level security is worth calling out, since a read-later library says a lot about a person and plenty of tools in this category treat that as an unstated default rather than a promise.

The product is in open beta right now and everything is free while that lasts. That covers unlimited archiving, unlimited bookmarks, AI summaries, deep search, the importer, Markdown export, highlights, notes, the Chrome clipper, and offline reading, all with no card required. When the beta ends, accounts move automatically to a standard free tier and existing saves are kept, though that permanent free plan is described as five saves, with paid tiers for more capacity not yet priced on the site. Beta terms like these usually signal a team buying usage and feedback rather than revenue, so the offer is generous now and the real cost arrives later.

That transition is the main thing to weigh before importing a large library. The commitment to keep existing data is stated plainly, and Markdown export means nothing is trapped either way, but anyone planning to save heavily should expect to be on a paid plan once the beta closes and should check the numbers when they're published. As a younger product it's also thinner on platform coverage than the incumbents, with a web app and a Chrome extension rather than native apps everywhere. For the core job, keeping a permanent readable copy of things worth keeping, it's already doing the work. Set against that, the archiving works today and the export path is real, which is the part that actually protects the hours spent building a library in the first place.