Read-later apps

Read-It-Later Apps Are Dead. Long Live Save-Anything Apps.

On November 12, 2025, Pocket permanently deleted every article its users had saved over 18 years. Anybody who missed the export deadline lost everything. Pocket was not a niche product. It had 20 million registered users and 2 billion saved articles at its peak. It was the biggest name the read-it-later category ever produced. And it is gone. Two years earlier, Omnivore shut down a month after ElevenLabs bought the team. Before that, del.icio.us, Google Reader, Readability, and StumbleUpon all met the same end. The read-it-later category has been dying batch by batch for two decades. The thing replacing it does not look much like what came before.

By Md Saban · Last updated: August 2026

A twenty-year graveyard

The read-it-later category started before it had a name. In 2003, del.icio.us moved bookmarks from the browser into the cloud and added tags. Users started tagging things "toread" to mark articles they wanted to come back to. The founder, Joshua Schachter, said people save a page for one of two reasons: they think they will need it again, or they do not have time now and want to read it later. That second reason spawned an entire product category.

In August 2007, a Minnesota developer named Nate Weiner built a Firefox extension and called it Read It Later. In January 2008, Marco Arment launched Instapaper so he could read articles on the New York subway without a signal. In 2009, Readability stripped ad-cluttered pages down to clean text. By 2011, Apple shipped Safari Reading List, a free built-in competitor that did less but was already installed. The dedicated tools kept building anyway.

Read It Later renamed itself Pocket in 2012, went free, and grew from 4.5 million users to over 20 million. Mozilla acquired it in 2017. Instapaper was sold to Betaworks in 2013, then to Pinterest in 2016. Readability shut down in 2016 after its revenue model failed. Google Reader shut down in 2013 and sparked an internet-wide panic about losing RSS. The 2020s brought a new wave: Matter raised $7 million, Readwise Reader entered public beta, and Omnivore open-sourced its entire stack. Then the shutdowns resumed. Omnivore closed in November 2024. Pocket closed in July 2025. Delicious had already closed in 2017 after five ownership changes.

The full ledger covers 13 major products launched since 2003. At least 8 are dead. Only Instapaper, Raindrop.io, Readwise Reader, wallabag, and Matter are still running in 2026. This is not a healthy category. It is a category where products die, users migrate, and the new product dies again a few years later. Something about the model is structurally broken.

Why read-it-later apps keep dying

The deaths look different from the outside but share the same causes underneath.

Saved, never read. Saving takes one tap. Reading takes 20 minutes. The ratio produces queues that grow until users feel guilty and uninstall. Research by Bergman, Whittaker, and Schooler found that 84% of bookmarks are never revisited after saving. Pocket's own data showed most saved articles had an effective lifespan of roughly 37 days before they were functionally forgotten. When users do not actually read what they save, they stop seeing the value and stop paying. Low retention kills the subscription math.

Free does not hold. Pocket traded revenue for scale in 2012, hoping a massive free user base would eventually monetize. It never did. Mozilla carried the cost for eight years and cut it in 2025 when budgets tightened. Omnivore was free and open source. Nobody paid the server bills. When ElevenLabs bought the team, the product vanished within a month. Free means someone else is paying. When they stop, the product stops with them.

Platforms give the basics away. Safari and Chrome both ship reading lists. WeChat has a floating window that turns any article into a bubble you can reopen anytime. Apple Notes and Google Keep both have web clippers. The core features of a read-it-later app, article extraction, offline reading, cross-device sync, are now free and preinstalled. To charge money, an independent tool has to be clearly better than the thing users already have. Most are not.

Acquisition usually means shutdown. Delicious changed hands five times and died. Omnivore lasted one month after acquisition. Pocket survived eight years under Mozilla and still got cut. The only counterexample in the entire 20-year history is Instapaper, which was bought by Pinterest in 2016, bought back by its operating team in 2018, and has run independently ever since. One survivor out of dozens of acquisitions across the category. When a big company buys a small tool, the product is usually dead within two years.

What changed between 2007 and 2026

Read-it-later apps were built for articles. The tools that replaced them were built for everything. The shift is not about better engineering. It is about what people actually save.

In 2007, the internet was mostly text. Blogs, news sites, forums, and long-form essays. Saving an article meant bookmarking a URL. In 2026, the average social media user has accounts on 6.52 platforms, according to DataReportal. Video now accounts for over 60% of total social media consumption. Instagram Reels reach 726 million people. The average TikTok user spends 70 minutes on the app daily. YouTube Shorts hit 70 billion daily views. The content people want to save is no longer mostly articles. It is short-form video, carousel posts, threads, newsletters, and memes.

The read-it-later model cannot handle this. Pocket was built to extract article text and display it in a clean reading view. It never understood Reels. It never indexed video transcripts. It never searched across platforms. The save button on Instagram and TikTok was built for engagement metrics, not for retrieval. Neither platform has a search bar inside saved posts as of 2026. Social media saves are useless without search. The old read-it-later tools and the new save buttons are both broken for different reasons.

How save-anything apps actually work

The shift from read-it-later to save-anything is a different approach to what happens after you save something. It is not a rebrand. It is a rewrite.

A read-it-later tool saves a URL and extracts the article text. A save-anything tool accepts content through the iPhone share sheet today, with Android access on the waitlist, and does far more on the backend. It transcribes video audio into searchable text. It reads the full article, caption, or thread body. It assigns topic tags automatically using AI classification, which can reduce manual filing. You should still review important saves because automated tags are not perfect. It generates a one-sentence summary so you can scan your library without opening everything. And it embeds all of that text into a semantic search index so you can find things by meaning. AI bookmarking understands meaning, not just keywords.

The retrieval experience is the biggest upgrade. In a read-it-later app, you search by title or URL and hope you remember the exact words. In a save-anything app, you search by describing what you remember. "That pasta recipe from a few weeks ago with the egg yolk technique" finds the carbonara Reel even when the caption was just "dinner tonight." Semantic search matches concepts across video transcripts, captions, and article text in one index. You stop needing to remember which platform something was on. You just describe it.

The typical person saving content across social platforms racks up roughly 3,650 saves per year at 10 saves per day. With traditional bookmarks, most of those would never be found again. With AI indexing, the save becomes genuinely retrievable. That is the difference between a graveyard and a library.

What the survivors have in common

The read-it-later and save-anything tools still running in 2026 follow the same playbook. It is the opposite of what the dead ones did.

Users pay directly. Instapaper charges $5.99 per month for Premium. Readwise Reader charges $9.99 per month. Raindrop.io runs on subscriptions from one developer. wallabag offers a hosted version at €11 per year. None of them are free, and none of them rely on venture capital or a parent company to cover costs. When users pay, the product answers to users. When investors or a parent company pay, the product answers to a spreadsheet.

The team stays small and cheap to run. Raindrop.io is one person. Instapaper is a small team. wallabag has a handful of contributors. Readwise Reader is the largest with a team in the low dozens. None of them need explosive growth to survive. A small team with paying users can run indefinitely. A large team burning venture capital has a countdown clock.

Data exports work anytime. Every surviving tool lets you export your data in a format other tools can read, usually HTML, CSV, or Markdown. Pocket users who trusted the product and missed the export deadline lost everything. Omnivore users got two weeks. The survivors treat data portability as a feature, not a concession. Your library is yours. You should be able to leave whenever you want.

There is an open-source or self-host fallback. wallabag and Slax Reader are fully open source. You can install them on your own server and the data lives with you. Even if the hosted version disappears, the code survives. That is not possible with most commercial tools, but the ones that offer it have a structural advantage: they cannot be shut down for you.

What the save-anything model still gets wrong

The new generation is better, but it is not a perfect fix. Several problems are unsolved.

AI tagging is good, not great. Even 92% accuracy means roughly 8 out of every 100 saves get the wrong tag. For casual content like recipes or productivity tips, that is fine. For niche technical content with domain-specific terminology, the error rate is higher. Most tools let you manually edit tags, but that reintroduces the manual work AI was supposed to eliminate. The gap between "good enough for browsing" and "reliable enough for a research library" is real.

Video transcription is imperfect. Auto-transcribing a Reel or TikTok with background music, multiple speakers, or heavy accents produces text that is searchable but messy. You can find the video, but the transcript might be garbled enough that you need to rewatch it to get the actual point. Transcription quality varies widely by language and audio clarity.

Most save-anything apps are new. TapFold, Fofo, Sprink, and Slax Reader all launched in the 2020s. They have not survived an economic downturn, an acquisition threat, or the kind of slow neglect that killed Pocket. The business model logic says subscription-funded tools with small teams are more durable, but logic and reality are different things. The survivors of 2036 will be the ones we cannot name yet because the test is time, not features.

None of them fix the saving reflex. The root cause of the read-it-later graveyard is not poor technology. It is the gap between the ease of saving and the effort of revisiting. Even the best save-anything app cannot make you actually engage with what you saved. A well-organized library of 5,000 unread items is still a graveyard. The tool can make retrieval frictionless, but it cannot make you open the app and search. The only fix for the saving reflex is behavioral: save less, save with intent, and review what you save.

Frequently asked questions

Are read-it-later apps actually dead?

The old guard is gone. Pocket shut down on July 8, 2025, and permanently deleted all user data on November 12, 2025. Omnivore closed in November 2024 after the team was acquired by ElevenLabs. del.icio.us died in 2017 after changing hands five times. Google Reader shut down in 2013. But the category is not literally dead. A new generation of tools has replaced it, tools that save more than articles. They save Reels, TikToks, LinkedIn posts, X threads, newsletters, and any URL. The name changed, but the problem is the same: you find something worth keeping and need a way to get back to it.

Why did Pocket shut down after 18 years?

Mozilla acquired Pocket in 2017 for an undisclosed amount and ran it for eight years. On May 22, 2025, Mozilla announced it was shutting Pocket down to focus resources on Firefox. The service stopped saving new articles on July 8, 2025, ran in export-only mode until November 12, 2025, and then permanently deleted all remaining user data. Mozilla's official statement said the way people save and consume web content had evolved. Pocket never found a sustainable business model on its own. It traded revenue for scale early, giving the product away free in 2012, and never reached a point where subscriptions could support the team without Mozilla's backing.

What is a save-anything app and how is it different from read-it-later?

A read-it-later app was built for one content type: web articles. It saved the URL, extracted the text, stripped the ads, and gave you a clean reading view. A save-anything app accepts any content from any platform. You share an Instagram Reel, a TikTok, a LinkedIn post, a YouTube video, a newsletter link, a Reddit thread, or a PDF into one library. The app reads the actual content, transcribes video audio if needed, tags it by topic, summarizes it, and makes it searchable by meaning rather than title. The design goal shifted from 'read this article later in a clean view' to 'get back to anything you saved, from anywhere, by searching what you remember about it.'

What is the best read-it-later app replacement in 2026?

It depends on what you save. If you mostly save articles and want a calm reading experience, Instapaper is the closest Pocket replacement and has survived since 2008 through three ownership changes. If you highlight and want those highlights to feed into a knowledge system, Readwise Reader is the power user pick at $9.99 per month. If you save across social media platforms and need unified search, save-anything apps like TapFold, Fofo, and Sprink cover Instagram, TikTok, YouTube, LinkedIn, and more. If you want full control and tech skills, self-host wallabag or Slax Reader. The best choice matches what you actually save, not what a comparison chart tells you to save.

What happened to all the articles people saved in Pocket?

Pocket users who exported before November 12, 2025 kept their data. Everyone else lost everything. Years of saved articles, highlights, tags, and reading lists were queued for permanent deletion. The export was an HTML file that could be imported into Instapaper, Raindrop.io, wallabag, or Readwise Reader. Pocket had roughly 20 million registered users and 2 billion saved articles at its peak. The scale of the data loss is unknowable because Mozilla never published final export completion rates. The lesson is not specific to Pocket. Any free service can shut down. Any service that does not let you export your data in a standard format is a time bomb dressed as a reading list.

Why do read-it-later apps keep failing while some survive?

The ones that die share a pattern: they are free, VC-funded, or owned by a larger company that eventually cuts them. Pocket was free and owned by Mozilla. Omnivore was free and open source. del.icio.us was free and changed hands five times. The ones still running in 2026 share the opposite pattern: users pay directly, the team stays small, and data exports are always available. Instapaper runs on subscriptions with a small team. Raindrop.io is one person funded by subscriptions. Readwise Reader charges $9.99 per month and took no venture capital. wallabag offers a hosted version at €11 per year and the code is open source. When users pay, the app answers to users. When investors or a parent company pay, the app answers to a spreadsheet.

Can I trust free save-later apps after what happened to Pocket?

You can use free tiers, but do not treat them as permanent storage. A free app that does not charge users is being paid for by someone else, and that someone else can stop paying at any time. If you use a free save-later app, check two things. First, can you export your data anytime in a format other tools can read, ideally HTML, CSV, or Markdown? Second, is the code open source so the tool can outlive the company? If the answer to both is no, treat the app as a temporary reading queue, not a library. The apps that survive decades are the ones where users pay enough to cover costs and exports are a feature, not a concession.