We’re finally shipping. It is hard to believe that in the age of AI coding, it took us nine months to build what is, on paper, a recording-and-transcription app.
I started Hemory to solve my own problem. Later I talked Jun, the CEO of WizNote, into joining. After ONES acquired WizNote years ago, he and I spent a lot of time talking about what a next-generation note product should be. He has spent more than twenty years in note-taking tools and knows the history of every one of them — Obsidian, Notion, and the rest go without saying. What I admire most is that he still uses Roam Research to this day (if you have not heard of it, that is fine; only the old hands know that thing). He threw himself into Hemory, coding until dawn night after night, and I had to keep telling him to go home and sleep. I would get up past eight and find that the code he committed at five had built green. His words: “Notes no longer exist.”
Back to my own need. Like everyone who wants recording and transcription, I started with meeting notes. There was pure software like Granola and Otter, and later recording hardware — Plaud’s industrial design is excellent, and I bought one the moment it came out. Once I had tasted what AI transcription could do, my standards went up. A few pain points these products could no longer meet.
First, it has to listen and keep a record continuously.
I always forget to start it, and forget to stop it. I would be halfway through a meeting before noticing it was never on; other times I would discover long after a meeting ended that it was still running, and several separate discussions had congealed into one mush. Between half and two-thirds of my working day is spent in meetings. Having to keep watching a recording app or device and operate it costs me an enormous amount of attention. I thought about just leaving it on, but most recording-and-transcription software caps a single file at four hours. And a single day’s topics can be scattered; I do not want them bluntly merged into one meeting-notes file.
Second, I want to retrieve what I said and heard along any dimension.
An instant memo the moment a meeting ends does not matter much to me. What I want is the chance to get something back when I need to remember it. And the way to get it back is not search in the traditional sense — I do not want to locate a few memos by keyword and then have to read them. No. Those memos have already lost fidelity. What I want are recollections about a given topic that cut across time and across meetings. With that memory in hand, I can do very rich downstream work with AI.
The experience I want is this: within hours I set in advance — say my workday, nine to eight — a service listens alongside me to everything that happens, and without any action from me it uploads, transcribes, and splits the day into separate meetings based on what was actually discussed. I can look at those segmented meetings and everyday conversations on a timeline to review what the day’s topics were, and I can read the summaries. Or I can use the agent I live in, Codex or OpenClaw, and talk to the service any time, mining and reworking what it heard of my work and my life.
I do not want meeting transcription. I want a tool that captures every conversation I have and automatically keeps it as memory.
I looked around. No such service existed. So we built one.
Hemory, from Hear and Memory. Hear it, and keep it.
The features are the ones I described above, all of them shipped. I have been building this product and using it at the same time. For more than seven months now, Hemory has been on by default whenever I am working. It is just software. It runs on iPhone, and it runs on Apple Watch. Hemory on Apple Watch gives you an experience no third-party recording hardware can match; I will write a separate piece about that.
The thing that surprised me most was connecting it to Codex and asking:
What interesting things did you hear yesterday?
What did you hear last month?
Write the PRD for the new features we have been discussing lately.
Write me a half-year review based on what you heard.
Turn the pricing strategy we have talked about several times this month into a full competitive analysis.
Watching Codex pull Hemory’s memory, think for a moment, and start producing the answer — my scalp tingled.
After that I never wanted to miss another day. I want all of it kept. A personal second brain, a digital assistant — without complete memory, none of it can know who I am, where I came from, or where I am going. Calling Hemory an assistant sells it short. Hemory is more like a clone of yourself. Once Hemory knows and understands everything I have expressed, and has heard every discussion between me and the people around me, I am him and he is me.
This is a small beginning. It sounds like a small product, but the few obsessive perfectionists on this team rewrote it three times before we dared to ship. Hemory’s mission is plain: to remember everything you said and heard, at the lowest possible cost.
“The limits of my language mean the limits of my world” — that one is Wittgenstein. Language is also how a person’s view of the world shows itself. AI keeps getting stronger; it can do anything, and it does more and more of it well. Born human, what we think and what we say have become the last thing that tells one of us from another. I asked Hemory to judge my MBTI from the way I express myself, based on what it had heard. The answer was uncannily accurate. My scalp tingled again.
I hope Hemory becomes a good companion for looking outward at the world and inward at ourselves — and, along the way, gets the work done for us.
Download it at Hemory.com. I hope you like it.