One Year in the App Graveyard
Start with the roll call, because the pattern only becomes visible when you see the cases side by side. These are not obscure tools that ran out of users. Several of them were loved right up to the end.
- Pocket Mozilla shut down the read-it-later service in July 2025, after nearly two decades of people saving articles into it. Users were given an export window, and according to the shutdown notices, data remaining after the October 2025 deadline was deleted. Millions of carefully curated reading libraries came down to one question: did you export in time?
- Notion Mail Notion has announced that its mail client shuts down on September 22, 2026, as the company refocuses on AI. Users are urged to export drafts, scheduled emails, snippets, and auto-label instructions by September 21. As of this writing, that deadline is weeks away.
- Samsung Notes sync The sync between Samsung Notes and Microsoft OneNote ended in July 2026. Nothing shut down in the dramatic sense, but new notes taken on a Samsung device stopped flowing outward, quietly stranding them inside Samsung's own apps. A bridge people relied on simply stopped existing.
- Rabrain The smaller second-brain tool went fully offline on August 1, 2026. To its credit, the team offered CSV and JSON export before the end, which is more grace than many shutdowns manage. Its users still had to notice the announcement, act on it in time, and then find a new home for what they exported.
- Limitless The AI wearable pendant stopped being sold after Meta acquired the company in December 2025, and reports at the time described service in some regions being cut off the same day the deal was announced. People who had bought a device to remember their lives for them learned how quickly a service can end when its parent company changes.
It matters that none of these were failures of love. Pocket in particular was adored, the kind of app people evangelized to friends and kept for fifteen years, and it died anyway. Rabrain was built by people who clearly cared, which is why they shipped export tools on the way out. Notion Mail is being retired by a healthy company making a strategic choice, not a desperate one. If quality or affection could keep software alive, most of these products would still be here. They are not, and that is the first honest thing to accept: the survival of an app has surprisingly little to do with how good it is or how much its users depend on it.
The second honest thing to accept is that this was an ordinary year. Nothing about 2025 and 2026 was unusually cruel to software. A comparable roll call could be written for any twelve-month window in the past two decades, and one will be writable for the next twelve. The graveyard is not a warning about a bad stretch. It is a description of the normal weather.
Why Shutdowns Are Structural, Not Scandalous
When an app dies, the coverage tends to reach for villains: greedy acquirers, distracted executives, betrayed users. The truth is less dramatic and more useful. Shutdowns are the ordinary tail of how software businesses work.
Look at the mechanisms behind the past year and you find three, none of them exotic. Acquisition: Limitless was bought, and the buyer wanted the team and the technology more than the standalone service. Pivot: Notion is concentrating its effort on AI, and a mail client that made sense in one strategy does not make sense in the next. Sunset: Pocket and Rabrain simply reached the end of what their models could sustain, at very different scales but for recognizably similar reasons. Acquisitions, pivots, and sunsets are not corporate misbehavior. They are the three doors through which nearly every product eventually exits, and companies walk through them because their obligations run to their own survival, not to your archive.
That last part deserves to be said plainly, without cynicism. A vendor's incentives end where your archive begins. While a product is alive, the company is paid to keep your data safe, and most take that duty seriously. The moment the product is scheduled to die, the economics invert: every month of continued data access is pure cost, every support ticket from a departing user is a distraction from whatever comes next. This is why export windows are measured in weeks rather than years. Pocket's users got months and a hard October deadline, after which the notices said remaining data would be deleted. Notion Mail's users are being asked to act by a specific day in September. Nobody involved is being villainous. The structure just stops paying anyone to care about your copy, and structures win.
Once you see shutdowns this way, the practical conclusion follows on its own. If the ordinary lifecycle of software includes death, and the notice period is a few weeks of emails you might miss, then the burden of surviving a shutdown cannot rest on the vendor's roadmap or your own vigilance. It has to rest somewhere structural too, and the only structure you control is where and how your data lives.
The Four Questions: A Data-Longevity Test
Here is the test. For any app that holds something you would mind losing, ask four questions. They take about ten minutes per app, and they tell you more about your real exposure than any amount of reading about the vendor's funding or roadmap.
Does a copy live on hardware you control?
This is the first question because it decides how much the other three matter. If a full copy of your data sits on your own disk, a shutdown is an inconvenience: the app dies, the files remain, and your problem is finding new software, not recovering your life. If the only copy lives on the vendor's servers, then everything depends on the export window, the export working, and you noticing the announcement in time. Pocket's deadline mattered so much precisely because for most users the service held the only copy. A local copy converts a countdown into a non-event.
Is it in formats other software can read?
A copy you cannot open is a souvenir, not a backup. Text, Markdown, PDF, CSV, JSON, and standard image and audio formats have each outlived every application that ever wrote them, and they will outlive whatever you are using now. A proprietary bundle that only the dead app could parse fails this question even if it sits safely on your own disk. When Rabrain closed, the CSV and JSON export was the difference between an archive and a brick: those formats meant the data could go somewhere else and become useful again.
Have you actually performed the export?
An untested export is a hope, not a plan. Export features sit in the least-visited corner of every app, and they break quietly: the button that times out on large libraries, the archive that omits attachments, the format that turns out to be a folder of HTML fragments with the images missing. The only way to know what an export actually contains is to run it once and open the result, and the middle of a shutdown window, with the support team already reassigned, is the worst possible moment for that discovery. Run the export on a calm day. Look inside the files. Then you know.
How many days' notice would you need?
The last question measures the gap between your situation and the warning you would realistically get. Shutdown notices arrive by email, often once, sometimes to an address you stopped checking. Limitless users in some regions reportedly had zero days between the acquisition announcement and losing service. Notion Mail users have weeks, which is generous by the standards of the genre, and still assumes they saw the news. If your honest answer is that you would need a month of lead time to notice, export, verify, and migrate, and the typical window is shorter than that, the gap is your risk, stated in days.
Grade every tool you depend on against these four questions and the results tend to sort cleanly. A tool that passes the first two barely needs the last two, because your data survives even a shutdown you never hear about. A tool that fails the first question is betting your archive on questions three and four going well, under time pressure, at a moment the vendor has stopped investing in your experience.
Why a Second Brain Deserves the Strictest Standard
The test applies to any app, but it does not apply equally, because data differs in how long it needs to live. A chat thread matters for a week. A project board matters for a quarter. A knowledge library is different in kind.
A second brain is the longest-lived data you own. The notes, saved articles, PDFs, and half-formed ideas you collect are meant to compound across decades: the point of writing something down in 2026 is to have it in 2036, connected to things you have not encountered yet. That time horizon changes the arithmetic of the graveyard. Over one year, the odds that any given healthy app disappears are low. Over the twenty or thirty years a knowledge library is supposed to serve you, the odds that at least one link in the chain breaks, an acquisition, a pivot, a sunset, a sync bridge quietly retired, approach certainty. You are not choosing an app for the next year. You are choosing the first custodian in a decades-long chain of custody.
This is why a cloud-only knowledge library fails the test before the test really begins. Whatever its features, it fails question one, and for data with a thirty-year horizon, failing question one means the library's lifespan is capped at the vendor's lifespan, minus the export window, minus your reaction time. People who kept years of reading in Pocket did nothing wrong by any ordinary standard. They simply held long-horizon data in a short-horizon arrangement, and the mismatch surfaced all at once, with a deadline attached.
So hold your second brain to the strictest version of the standard: a local copy on your own machine, in formats other software can read, verified by an export you have actually opened, with a notice tolerance of zero, meaning the app could vanish tomorrow, unannounced, and you would lose nothing but the interface. That is not paranoia. It is matching the durability of the container to the lifespan of what it holds.
The Honest Caveat: This Test Applies to Mindly Too
An article like this, on the website of a note-taking app, has an obvious problem: every vendor says their product is different, and the graveyard is full of products whose vendors said exactly that. So let us apply the test to Mindly, plainly, as a vulnerability rather than a boast.
Mindly is a company, and any company can disappear. We do not plan to, but neither did anyone else on the roll call above, and you have no reason to weigh our intentions more heavily than Pocket's or Rabrain's. That is precisely why Mindly is built so that our intentions do not matter to your archive. Your library lives in a folder on your Mac, not on our servers, so a copy sits on hardware you control by default rather than by discipline. You can export it to standard formats other software can read. Run that export today if you like, and open the files, so question three is answered by evidence instead of trust. If Mindly vanished tomorrow, suddenly and without notice, your files would still be yours: readable, on your own disk, ready for whatever you use next. On the four questions, your notice tolerance is zero, and it stays zero no matter what happens to us.
The same logic shapes smaller decisions. AI processing happens over encrypted channels and your content is not retained after the request is served, so there is no second copy of your library accumulating on someone else's infrastructure. And if a Pro subscription lapses, items beyond the free limit become read-only rather than deleted, because a lapsed payment should never cost you data. None of this makes Mindly immortal. It makes Mindly survivable, which the graveyard suggests is the only promise worth listening to.
A Practical Checklist for This Week
Knowing the test is not the same as having run it, and September 2026 makes the difference concrete: if you use Notion Mail, an export deadline is weeks away as this is published. Here is the whole practice, compressed into an hour or two.
- List what you would mind losing Not every app, just the ones holding data you would grieve: notes, saved reading, documents, photos of whiteboards, voice memos. For most people it is three to six apps.
- Act on live deadlines first If any app on your list has an announced shutdown, stop and do that export now. Notion Mail users have until September 21 to export drafts, scheduled emails, snippets, and auto-label instructions. Announced windows do not reopen.
- Run the four questions on each app Local copy, open formats, tested export, notice tolerance. Write the answers down. The apps that fail question one are your real exposure, whatever their roadmaps look like.
- Perform one real export per app Today, not during a shutdown. Open the result and check that attachments, images, and metadata actually made it. You are converting hopes into facts.
- Put a repeat on the calendar A quarterly export of anything cloud-only turns the worst case from losing years into losing weeks. If everything important already lives locally in open formats, this step takes minutes.
- Move long-horizon data to a survivable home For data meant to last decades, your knowledge library above all, choose an arrangement that passes all four questions by design, so the practice stops depending on your vigilance.
That last step is where Mindly fits, and the fit is specifically about longevity rather than features. You capture anything with one shortcut, ⌘M, and notes, links, PDFs, files, screenshots, and voice memos land in a library that lives in a folder on your own Mac. The AI does the organizing that usually decays first in a migration, tagging, summarizing, transcribing voice, reading text inside images, and linking related items, and you find things again by searching in plain language, chatting with any saved item, or seeing the connections laid out as a mind map. Underneath all of it sit your files, on your disk, exportable to standard formats, which means the second brain you build passes the test on the day you start it. The free tier holds 25 items with no account needed, and Pro is €7.99 a month or €44.99 a year.
The apps in the graveyard were not mistakes, and using them was not foolish. People got real value from Pocket for years, and the readers who exported in time lost nothing but an interface they were fond of. That is the whole lesson, and it is an oddly hopeful one. Apps die, and you cannot change that. Whether your data dies with them was always the part you control.
Free for macOS, no account needed. Build your library in a folder on your own Mac, where no shutdown can reach it. Download Mindly →