The Promise: Perfect Memory as a Product
Before getting critical, the category deserves honest credit, because the engineering pitch was mostly delivered. These devices do what they say they do, and what they do is genuinely new.
The recorders vary in form, a pendant on a magnetic clasp, a pin on the shirt, a small disc on the wrist, but the loop is the same. The device listens through the day, syncs audio to a companion app, and the app turns it into text. From there the software layers begin: continuous transcripts of everything said around you, daily summaries that compress hours into a page, to-dos and reminders extracted from things you said you would do. Ask the app what your colleague said about the deadline and it can often tell you. As of 2026, Plaud's NotePin line is widely treated as the leader of the category among devices still on sale, and the reporting tends to agree on the fundamentals: battery that lasts a working day, transcription accurate enough to trust, summaries that read well.
- Continuous transcripts Everything said around you through the day, converted to text you can scroll and search, with the audio kept underneath in case the words matter exactly.
- Daily summaries A compressed account of the day, the meetings, the themes, the notable moments, generated without you asking for it.
- Extracted commitments The things you said you would do, pulled out of conversation and turned into reminders, which is quietly one of the most useful tricks in the category.
- Ask-your-day answers A chat interface over your recorded life, so a question about what was said can sometimes replace scrolling entirely.
It is worth being precise about what this genuinely fixes, because some of the oldest failures of note-taking simply disappear. You no longer have to choose between participating in a meeting and documenting it. The idea that arrives on a walk, historically the most-lost artifact in knowledge work, now survives. The verbal agreement that both parties remember differently three weeks later has a record. For lectures, interviews, and long working sessions, a reliable transcript is an unambiguous good, and pretending otherwise would be dishonest.
The capture problem, the physical difficulty of getting spoken information into text, is done. But solving capture does not solve memory, and the gap between those two things is where the rest of this article lives.
The Transcript Landfill
Here is the uncomfortable arithmetic of recording everything. A single workday of always-on capture produces hours of audio and tens of thousands of words of transcript. A month produces a small library. Nobody reads a small library every month.
This pattern has a familiar shape, because most of us have already lived through it once with read-it-later apps. Saving an article feels like progress, so people save hundreds, and the queue quietly becomes a landfill: a growing archive that generates guilt instead of knowledge, until one day it gets declared bankrupt and deleted. Transcript archives fail the same way, for the same reason. The act of recording feels like remembering, so the discipline that would actually produce remembering, reviewing, distilling, filing, never happens. The archive grows. The knowledge does not.
Recording is capture without curation. A transcript is raw material, and raw material that never gets refined is not knowledge. It is inventory.
The test that matters is not whether the recording exists but whether you can answer a question with it three months later. What did we actually agree about the launch scope in that call in April? The raw transcript technically contains the answer, but finding it means remembering which day the call happened, opening that day, and reading. Keyword search helps only if you can guess the exact words someone used, which is precisely the thing you offloaded to the machine. The daily summaries help more, but a summary written by a model that did not know what would matter to you in three months tends to flatten the one exchange you need into a line about discussing project timelines. Volume does not just fail to help retrieval. Past a certain point it actively defeats it, because every additional recorded hour makes the needle proportionally smaller.
It is tempting to answer that the summaries and the ask-your-day chat are the curation, and they do help. But a generated summary is a lossy average of a day, not a judgment about what you will need, and a chat over your archive is only as good as the question you remember to ask it. Curation, in the sense that matters, is a decision: this exchange was important, this idea belongs next to that project, this quote should surface when I am writing the report. Software can assist that decision and file its results, but the archive that skips the decision entirely just becomes a better-indexed landfill.
The Limitless Parable
The second lesson of the always-on era did not come from a productivity essay. It came from the news.
Limitless made arguably the best-known pendant in the category, and the people who wore one were its true believers, people who had committed months or years of their conversations to it. Then, as reported, Meta acquired Limitless in December 2025. The pendant stopped being sold to new customers, and on the day the acquisition closed, service in the EU, the UK, and several other regions was cut off, a move widely attributed to the stricter privacy regulation in those markets. Owners there were left holding a device whose cloud had gone quiet.
It is worth telling this story fairly, because the tempting version, big company kills beloved product, misses the useful lesson. Meta made a business decision of a kind that happens constantly; acquisitions reshape product lines every year, and no malice is required. The fragility was structural, and it existed long before the acquisition: an archive of irreplaceable personal recordings, your meetings, your conversations, pieces of your family's life, held in one vendor's cloud, readable only through one vendor's app, subject to one vendor's strategy. When custody is that concentrated, someone else's quarterly planning can switch your memory off, and there is nothing unusual about the company that does it.
The rule this teaches is older than wearables: data that matters to you should survive the death of the app that created it. A transcript that exists only inside a vendor's cloud is a loan, not a possession. Export while exporting is possible, keep the copy somewhere you control, and treat any tool without a real export path as a place you visit, not a place you live.
What a Working Pipeline Looks Like
The two failure modes, the landfill and the lockout, point at the same fix from different directions. What survives is a pipeline with three short stages: record deliberately, distill to what mattered, and give the result a permanent home.
Record where recording genuinely helps
The always-on pitch is total capture, but the value is not evenly distributed across your day. It concentrates in a few situations: meetings where decisions get made, lectures and interviews, working sessions where ideas move fast, and the thinking you do out loud on a walk. Recording those deliberately, starting a recording when something is worth keeping, costs you a little coverage and buys you an archive made mostly of signal. If you do wear a recorder all day, the same principle applies one step later: treat the continuous transcript as a quarry rather than a library, and pull out the blocks worth keeping while you still remember why they mattered.
Distill while it is fresh
Distillation is the step the devices cannot do for you, because it requires knowing what you will care about later. It is also mercifully small. For most recordings, what deserves to survive is a few sentences: the decision made, the action items and who owns them, the idea worth developing, the quote worth keeping exactly. Write that down, or take the machine summary and correct it while the conversation is still warm. The transcript itself does not get thrown away; it gets demoted to what it really is, source material that stays attached to the distillation in case you ever need the exact words.
Move both into a library that retrieves
The last stage decides whether any of this compounds. The distillation and the transcript need to land in the same place as the rest of what you know, next to your notes, your PDFs, your saved links, where one search covers everything and related material can connect. Voice deserves the same treatment as every other format. The moment recordings live in their own silo, a separate app with its own search that covers nothing else, you have rebuilt the vendor-cloud problem one layer up, and guaranteed that the answer to any real question is split across two systems. And, given the Limitless lesson, that library should sit somewhere you own, on hardware you control, in a tool whose disappearance would not take your archive with it.
Written out like this, the pipeline can sound like homework, so it is worth saying what it costs in practice: a minute or two after each recording that mattered, and nothing after the ones that did not. That is the whole discipline. It is dramatically less work than the read-everything ambition the always-on pitch quietly implies, and unlike that ambition, it compounds, because every distilled recording makes the library slightly better at answering the next question.
The Consent Note
There is one more cost of always-on recording that has nothing to do with retrieval, and it is the one most coverage mentions with visible discomfort: everyone else in the room.
A wearable recorder does not just capture your day; it captures everyone who talks to you, mostly without their knowledge. The commonly reported experience is that people who notice the device become guarded, and people who learn about it afterward feel ambushed, which is a real social price before any law enters the picture. And law does enter the picture: recording and consent rules differ sharply by region, some places require the consent of everyone in a conversation, and nothing here is legal advice, so if you record people, knowing your local rules is on you. The etiquette is genuinely unsettled and will stay that way for a while. It is worth noticing that deliberate recording largely dissolves the problem: when you record specific meetings rather than your whole life, asking the room for permission takes one sentence, and everyone gets to know which version of you they are talking to.
Where Mindly Fits
Mindly is a second brain app for macOS, built around the part of the pipeline the recorders leave open. Press ⌘M anywhere on your Mac and speak, and the voice memo is transcribed, summarized, and tagged by topic automatically, landing next to your notes, PDFs, links, and screenshots rather than in a separate voice silo. The idea from the walk becomes a searchable note with no filing step. If you already use a wearable recorder, its output has a home too: drop transcript files into the library and they are treated like everything else, tagged, connected to related items, and covered by the same search.
Retrieval is the point of all of it. Search works in plain language across every format at once, so what did we decide about the launch scope is a query rather than an archaeology project, and you can open any saved item, a transcript included, and chat with it to pull out the decision or the detail you need. A mind map view shows how the things you have saved connect. And the library lives locally on your Mac, which is the quiet answer to the Limitless problem: no vendor cloud holds your archive, so no acquisition can turn it off. AI processing runs over encrypted channels, and your content is not retained after the request is processed.
The free tier holds up to 25 items with no account needed, enough to run the loop on a real week of ideas and meetings. Mindly Pro, at €7.99 a month or €44.99 a year, removes the limit and includes advanced voice transcription, so the capture half of your pipeline is covered on the days you leave the wearable at home.
Capture is solved. Give the part that matters a permanent home: voice memos, transcripts, notes, and files, organized automatically and searchable in plain language, in a library that lives on your Mac. Download Mindly →