The Graveyard
A week spent reading obituaries of minimal capture tools. The survivors invested in the exit, not the entrance, and the strongest evidence against my own rules came from KakaoTalk.
Date
Category
Writer
I spent a week reading obituaries.
Not out of pessimism. Out of method. If you are building in a category, the single most valuable evidence available to you is what happened to everyone who already tried. It is free, it is real, and unlike a user interview nobody was being polite to you.
What follows is a survey of minimal capture tools and what became of them, and the pattern that emerged. I have tried to be accurate about status rather than dramatic, and there is one correction to my own earlier notes in here.
Fig. 1. The catalogue room of the Bibliothèque Mazarine, Paris. Some archives last three hundred years. Almost none of them are products. Photograph by Marie-Lan Nguyen, CC BY 2.0 FR, via Wikimedia Commons.
The roll
Product | Status | What happened |
|---|---|---|
Simplenote | Unmaintained, still running | Acquired 2013 with an explicit promise of active development. By 2025 the position was that it "is no longer in active development" |
Mem | Repositioned | Raised around $23.5M, moved toward an AI team workspace |
Tana | Repositioned | Raised around $25M, moved toward an enterprise AI knowledge graph |
Heptabase | Repositioned | Began as visual note-taking, moved toward AI-assisted learning |
Roam Research | Stalled | Defined a moment. Its own advocates now describe their graphs as accumulations nobody returns to |
Reflect | Alive, deliberately small | Roughly 1,371 paying customers, healthy specifically by declining to become something else |
Drafts | Alive, ten years | Same pitch since launch. Survives on a layer almost nobody copies |
Table 1. Minimal capture tools and their outcomes, as of September 2026.
A correction I owe on Simplenote
I had this filed in my own notes as a hard shutdown in July 2025. That is wrong, and I want to correct it in the same place I would have repeated it.
Simplenote was not shut down. Automattic's position, stated in 2025 support communication, is that it "is no longer in active development, and while the app remains available, only its basic functionality is being maintained at this time."
The distinction matters, and not only for accuracy. A shutdown is a decision. Unmaintained is a slow outcome, and for the argument I am making it is the more interesting of the two. Nobody killed Simplenote. It simply stopped being worth anyone's attention while continuing to work perfectly, which is a harder fate to design against.
The original promise, from the 2013 acquisition announcement, reads differently now: "You know how sometimes, the services you love just disappear when they're bought by someone else? Or they wither and die a slow and painful death? Not the case here."
The pattern, and the exception that explains it
Almost every survivor in that list changed what it was for. The ones that pivoted moved toward AI, toward teams, toward the enterprise. That is, toward a buyer, and away from an individual with a thought.
This is not a moral failing. It is a reasonable response to a real economic fact: an individual with a thought is a very small amount of money, and an enterprise knowledge graph is a large one. If you have raised twenty-five million dollars, the first market cannot service the obligation and the second can.
The exception is Drafts, and its exception is the entire finding.
Drafts does not survive on capture. Capture is the easy half and everyone has it. Drafts survives on its Actions layer: the machinery that gets text back out and into somewhere that it does something. Into a message, a task, a document, an append to an existing file, a shortcut, a script.
The products that lasted invested in the exit, not the entrance.
Ten years is a long time to be right about something this unfashionable. Every competitor spent its design budget on the moment text arrives. Drafts spent its budget on the moment text leaves, which is less demonstrable, harder to screenshot, and turns out to be where the value was.
The two things that beat all of them, and neither is a product
Now the uncomfortable part.
The most successful instances of "just drop it somewhere" are not products at all. A system quick-note gesture. Messaging yourself in a chat app.
They did not win on design. They won on adjacency. They are free to reach because you are already there.
And share-sheet position, which looks like the one dimension you could compete on, is not adjustable. On both major mobile platforms the ordering emerges from repeated use. First position is an outcome of habit, never a setting. You cannot design your way into it, you cannot buy it, and you cannot ship a feature that grants it.
This is the structural disadvantage of every standalone tool in this category, and I have not found anyone who has solved it. The honest version of my own position is that I have not solved it either. What I have is one observation that runs the other way, which I will come to.
The detail I keep returning to
Of everything in a week of reading, this is the one that stuck, and it is not from a startup.
The most successful "drop it in one place" behaviour in Korea is messaging yourself on KakaoTalk. Enormous adoption, zero marketing, no product team assigned to it. It is simply a chat with yourself.
In 2019 Kakao shipped a 서랍, a drawer, for that chat: the ability to categorise, star and filter what had accumulated there.
Read that again. The single most successful instance of undifferentiated accumulation was, by its own maker, upgraded away from undifferentiated accumulation. Not because a competitor forced it. Because the pile had become hard to manage and the people using it asked.
That is the strongest available evidence against my own no-organisation rule, and I am not going to file it somewhere I will not see it again.
My position, stated so it can be held against me
The rule survives at current scale, because the closest research suggests recall is near-perfect with no organisation at all when a collection is small. Bernstein and colleagues found one unintelligible scrap in 533.
It should be reconsidered at exactly one trigger: when a real history has grown large enough that the absence causes a repeated, real failure. Not before. Not because a competitor has folders. Not because a reviewer asked.
And if that day comes, the right response is probably not folders. Kakao's drawer arrived years after the behaviour did, which means the behaviour came first and the structure was fitted to it. That is the correct order, and it is the opposite of how most of us build: we ship the structure on day one and then hope a behaviour arrives to fill it.
What a graveyard is actually for
A week of this is genuinely depressing and I would recommend it to anyone.
The value is not in the list of failures. It is that the list makes the shape of the category visible in a way no amount of thinking about your own product will. Seven companies, hundreds of person-years, tens of millions of dollars, and they converge on the same two destinations: repositioning toward a buyer, or staying small on purpose.
There is no third outcome in the data. Which means the strategy question is not "how do I build a better one." It is "which of those two am I choosing," and the honest answer for a studio building this alongside client work is the second one, chosen deliberately rather than arrived at by failing to reach the first.
Sources
Catalogue room, Bibliothèque Mazarine, Paris. Photograph by Marie-Lan Nguyen, CC BY 2.0 FR, via Wikimedia Commons. commons.wikimedia.org/wiki/File:Salle_des_catalogues_de_la_Bibliotheque_Mazarine_Paris.jpg
Bernstein, M., Van Kleek, M., Karger, D., & schraefel, m.c. (2008). Information scraps: How and why information eludes our personal information management tools. ACM Transactions on Information Systems, 26(4). DOI 10.1145/1402256.1402263
Related reading
Latest Articles.
© Moc Mallen Studio
Thoughts, ideas, and perspectives on design, simplicity, and creative process.
