Launching today
Auday
Turn your Apple Watch into your all-day AI recorder.
7 followers
Turn your Apple Watch into your all-day AI recorder.
7 followers
The best AI recording hardware? You may already own it. Auday turns Apple Watch into your all-day AI recorder. Beyond voice, Auday combines health, real-time stress, location, and other signals to understand your everyday life, not just answer your questions. Almost everything runs locally on your devices. Configure your own API key for AI features—we don’t even have a backend server. $9.90 as a one-time purchase. No subscription.
Payment Required
Launch Team
TinyFishSearch, fetch and browse for AI agents. Claim 30% bonus!
Promoted
Hi everyone,
The best AI recording hardware? You may already own it.
I'm excited to share the story behind Auday and the thinking that went into building it.
1. After trying so many AI recorders, I ended up choosing Apple Watch
1.1 I tried several devices, but none quite felt right
Over the past year, I bought at least five AI recording devices. Some were names you'd probably recognize; others were more niche or experimental. An AI recorder is a little different from a traditional voice recorder. Beyond capturing sound, it can summarize it automatically and use AI to help you organize what you've recorded. There's a real reason this category has taken off: it meets a real need. A lot of us, myself included, record things as a kind of insurance. We might not listen back right away, but when we need a recording, we want it to be there. The trouble is, if you record everything as you go, you eventually end up with hundreds of files you can't find your way around. The advantage of AI recording is that it can make sense of that mess. For people who forget things easily or whose attention is constantly interrupted, including people with ADHD, that help can make a huge difference.
So I spent the past year exploring these products, looking for something that really suited me and that I could see myself using for a long time. The first one I bought was an AI recording card that attaches to the back of your phone. I picked a model that was very popular at the time. But with its leather sleeve, it was still a little thick, and I couldn't get used to how it made my phone feel. Later, I went looking specifically for a very thin one, among the thinnest available at the time. Even then, holding a phone, a phone case, and the recording card all together still felt uncomfortable. The card was already impressively thin, but when you use your phone as often as I do, even that little bit of extra thickness matters.
Then there was charging. I already had to charge my phone, iPad, and watch every day. Adding another thing attached to the back of my phone just wasn't very convenient. As a result, I barely used it. Usually, I'd only bring it along and attach it before an important meeting. Later, I tried small wearable recorders and some more stylish AI recording devices designed to look like accessories. But I kept coming back to the same questions: How good is the recording? How easy is it to carry? Does it feel out of place when I wear it? And does it feel like one more device I have to deal with? That burden is both physical and mental. And, of course, there was the most important question of all: privacy.
Despite some disappointing experiences, I never lost my enthusiasm for AI recording hardware. I had so many ideas about what it could become. Today's AI agents and agent harnesses can already turn our computer activity, online activity, and the information on our screens into context. Their connection to real life, though, is still much weaker. Conversations in the real world reach an actual person first, who then has to relay them to the AI on a screen. But what if AI could remember those conversations directly? Would I still have to keep explaining everything myself? Suddenly, it would have much more background to work with and could help me in many more ways. That's why I've kept believing in AI recording.
The companies making AI recorders are exploring this direction too. Everyone wants longer battery life, better audio capture, and more capable AI. Some products have already started exploring all-day recording or integrating more powerful agents so they can work with the AI on our screens. But there's a tension here: the more you record, the more you worry about privacy. Uploading the occasional meeting feels completely different from handing over all twenty-four hours of your day. And with so many startups entering this space, there's no shortage of different approaches to privacy. But honestly, when it comes to trusting someone with my entire day, I still worry. I kept looking, but never found an option I felt completely comfortable with.
1.2 Discovering what Apple Watch could do
Then one morning, after waking up, I checked the sleep data on my Apple Watch as usual and suddenly realized something: apart from my phone, wasn't this the one piece of hardware I already had with me almost every day, from morning to night? I'd been using this Apple Watch for a while. Because I like exercising, wearing it and charging it had already become as natural as using my phone. I'd been searching for a device that wouldn't add another burden, and it was already on my wrist. At that point, it was only a passing thought. Later, though, I saw people on social media using Apple Watch to record concerts, and that really set something off.
That was when I discovered how many people had already made Apple Watch their first choice for recording concerts because of its audio quality. One thing kept coming up in the posts I read: “Listening back feels like being there again.” After that, I couldn't stop thinking about turning Apple Watch into an AI recorder. I already wore it every day, it recorded well, and I wouldn't need to buy or carry another device. It felt like all the pieces were there. Why not give it a try?
The more I researched, the more convinced I became that Apple Watch was the best hardware for this right now. After so many years of refinement by Apple, it has become one of the company's most mature product lines. Its hardware, battery life, sensors, and the way its software and hardware work together are already exceptionally well developed. That's years of work a startup would struggle to catch up with in a short time. After some initial real-world battery tests, I felt even more confident about making it an all-day AI recorder.
Most importantly, Apple Watch can do much more than record audio. Heart rate, sleep, exercise, the real-time stress readings I particularly like checking, and more than eighty types of workout tracking can all fit into this tiny watch. One day, I tried yoga with my family, and it detected it straight away. In that moment, I was honestly impressed by how good its detection was. When you wear it every day, there may be plenty of features you've barely paid attention to. It was only when I started developing around it that I realized just how much was already packed into this watch.
Add location, and those sensors already have the potential to preserve your real life, your physical condition—who you are—as context. By that point, I was incredibly excited. I'd started out wanting to build an AI recorder. Now, I wanted to build an AI recorder that could capture the context of a life. Your daily life, your body, your health, and your exercise could all connect with your recordings. In turn, those signals could help AI understand the recordings better.
(Continued in the next reply — part 2/4.)
Product Story — continued (2/4)
For example, a traditional AI recorder mainly turns audio into text and then asks AI to find the important parts. But your physical state can be a direct clue too. If my stress level suddenly rises during a certain period, the conversation from that time might deserve a closer look. Maybe I'm in an important meeting, or maybe I've just received an important phone call. Or imagine my heart rate is noticeably different from usual. The AI also sees that I'm talking about life with someone I have a crush on, and my location has changed from the office to a beautiful restaurant. Put those details together, and it has a much richer understanding of what I'm experiencing. It could easily connect those little details with a lovely date.
The more I explored these possibilities, the more excited I became. Apple Watch wasn't just capable of doing the recording well. It gave me a chance to connect a person's real-world state with their recordings. Information that had been scattered across different places could work together here. That went well beyond what I'd originally imagined for AI recording hardware.
There was another surprise. One reason I'd attached a recording card to the back of my phone was to capture some phone calls. Those cards often advertise vibration microphones: they sit against the phone and pick up sound through its vibrations. That was a small but real need for me. In practice, though, the results were sometimes frustrating. The original audio was almost impossible to make out, and I was mainly relying on the noise reduction and transcription afterward to recover the conversation. So when I first chose Apple Watch, I thought this might be the one thing it couldn't cover. Then, during a development test, I answered a call while wearing the watch and holding the phone in that same hand. Later, when I happened to play back the recording, I discovered that I could hear the call too. I was thrilled. The watch was on my wrist, so close to the phone in my hand, that its position had actually become an advantage for picking up sound. Something I'd thought required a recording card had, this time, worked with the watch as well. At that point, I couldn't see a reason not to keep building.
1.3 After the launch event, I picked up the pace
I'd actually had the idea back in May. But with so much going on in real life, I kept putting it off. Then September came, and Apple held this year's launch event. As usual, I was excited to see the new iPhoneDuo and all the other new products. But when Apple Watch Series 12 appeared, I suddenly couldn't sit still. Apple announced its new Audio Intelligence features. Among them was Live Rewind, planned for release later in the year, which could show the last fifteen seconds of a conversation as text so you could revisit something you'd missed. At the same time, heart-rate and HRV detection had also received substantial improvements. I'd been researching how to use that physical data to add context to recordings, so seeing Apple push the hardware further was incredibly exciting. But it also made me tense. It seemed Apple had realized the same thing: Apple Watch could become AI recording hardware, and a very good choice at that. On one hand, I was delighted that an idea I'd had in May had led me in the same direction as Apple. On the other, I was nervous. I'd had the idea for months, but my actual development progress felt like it was only at ten percent. After the event, the anxiety really hit me: I couldn't keep putting this off. I had to make it happen.
The funny thing was that Apple's post was titled “Stay up late tonight, or get up early?” And yes, after watching the event, between the excitement and the anxiety, I barely slept that night. Haha.
So after the launch event on September 10, I worked on it for a full week straight. More development, testing, and revisions followed, until it finally became the Auday I'm sharing with you today. As someone who has used AI recording hardware a lot, and talked with many other people about their experiences, I wanted first to make a product I was happy with myself. I also wanted to resolve as many of the frustrations people had with these devices as I could. I've spent this whole first section telling you how I got here. Now I can finally get into what Auday actually does, and why I think it's worth trying.
2. What Auday does, and why I built it this way
Let me introduce Auday in two sentences. It's software that turns your Apple Watch into an all-day AI recording device: the core functions—recording, transcription, and search—run locally on your devices, while the additional AI features use an API key you configure yourself. It's a one-time purchase at a very low price, with no Auday subscription, no new account to register, and not even a backend server on our side. Most of the choices below came from my own experiences using AI recording hardware over the past year.
2.1 Privacy
For an app that I hope will stay with you for long stretches of your life, Auday records more than sound. It can also connect location, physical state, activity, and health information. There's so much private information involved that privacy has to be a top priority. It's also central to whether anyone can trust software made by a small developer like me. Honestly, if I built a cloud service and tried to compete with the big companies on security, I couldn't confidently say I'd do a better job. But I couldn't ignore the problem either. So I chose to do as much of the processing as possible on the devices themselves.
Apple Watch transfers recordings to its paired iPhone through Apple's WatchConnectivity framework. It doesn't rely on iCloud, and there's no need to upload the recordings to an Auday server first. Transcription on iPhone uses a local model, and search uses a local semantic embedding model too. Recording, transcription, reading, searching, and listening back can all happen on your own devices. Even without configuring any cloud AI, that already covers the basic needs of a tool for recording over long periods.
To go a step further, I'll be honest: I also really wanted to build an AI subscription business. But the approach I had in mind would have required my own account system, connecting users to AI on their behalf, and routing their content through my service. That raised two kinds of concerns: users' accounts and identity information, and the private content they would send over for AI analysis. Even if I promised, as some companies do, to delete everything immediately after processing and never retain your data, I think people would still worry. I wanted to use this myself for a long time, so I completely understood that concern.
In the end, I took out both the account system and the server that would relay users' content. You buy the local, private version once and own it permanently; for AI, you configure your own API key. You can connect a model you've deployed yourself, or choose a major model provider you trust, with the iPhone connecting directly to it. When you want summaries, highlights, daily reflections, or Chat, the relevant text and accompanying context go directly to your chosen provider. The original audio stays on your devices. This decision certainly means giving up a lot of potential revenue, but for a product like this, I think it's worth it.
(Continued in the next reply — part 3/4.)
Product Story — continued (3/4)
2.2 Buy once, no subscription
The first reason for a one-time purchase was the privacy consideration I just described. Beyond that, I've always felt that hardware as good as Apple Watch shouldn't be limited to the uses we've already grown accustomed to. People sometimes jokingly call it a “beautiful waste of money.” Some buy one and leave it gathering dust; others only occasionally check their health data. But this is something Apple has spent years refining, in both hardware and software. There's still so much potential to explore. Once I realized how well suited it was to AI recording, I really wanted to make that accessible to people who already owned a watch, at a very low cost.
Auday's regular price is $19.80. For the launch, it's half price: $9.99 as a one-time purchase, and I plan to keep this promotional price for quite a while. A friend saw the price and bluntly told me I wouldn't even earn back what I spend on the AI subscriptions I use to write the code. My answer was, “I don't care.” Realizing that Apple Watch had the potential to become an excellent AI recording device was exciting enough that I wanted to share that discovery with everyone.
2.3 The different ways AI fits into Auday
Auday's AI features mainly focus on three areas. The first is organizing hourly and daily transcripts, and automatically finding highlights. As in the examples in the first section, it can draw on more than the words transcribed from audio: it can also use the physical state, activity, and location information you choose to connect. That gives it a fuller picture when looking for things worth paying attention to. You can also mark a moment manually on the watch. If something feels important, just tap the screen to give the AI a signal to pay closer attention to the conversation around that time. Then you can carry on with what you're doing and come back later to the key points and hourly notes it has put together.
The second is daily summaries and reflections. As an observer that stays with you throughout the day, Auday can bring together what you say and do, your physical state, and your experiences, then offer an outside perspective on things you might otherwise overlook. It can help you reflect on yourself and find things to adjust or improve. This is something I'm especially excited about: recording life, while also helping myself keep growing.
The third is Auday AI Chat. I don't want to simply add a question-and-answer assistant, as so many AI apps do, and stop there. We have more than audio—we have this rich context from your life, and those records will keep accumulating. It can become something much more personal. People already use lots of AI assistants and agents. I want Auday AI to be the most intimate, the most private, and the one that knows you best. It has its own memory system: from hourly and daily conversations and reviews, it extracts lasting things that are truly specific to you, including your preferences, habits, important relationships, and ongoing concerns. It stores them locally as long-term personal memories and brings them into future reviews and conversations. Its understanding of you doesn't start over every day; it keeps building. The longer you use Auday AI, the better it knows you.
I especially like explaining this through dating. Imagine you ask Auday AI, “Do Emma and I really like each other?” It can consider much more than what you said over that one dinner. For example, it can look at your physical state when you're together alongside how you usually feel at work. You might often be tense while working, yet much more relaxed around her. Beyond what you said to her in person, your reactions after the date and the little things you couldn't help saying to yourself on the way home could also provide clues. Because Auday records over long periods and stays with you, it can put things that happened at different times into the same context.
With an agent architecture and context retrieval, it can look further back too. Say your records show that, in the week before your first date, you suddenly became very serious about working out and kept mentioning your physique. It has a chance to connect those things: were you trying to get yourself into better shape quickly, hoping to make a good first impression on her? You might simply ask, “Do I like her?” But it can talk that through with you using experiences you may never have consciously connected yourself. I find that kind of experience fascinating, and it's something I'm really looking forward to.
So Auday AI starts with AI recording, but what I want it to do goes far beyond recording itself. Sound, physical state, location, the conversations before and after a moment, and personal memories accumulated over time can all help it understand you. It can be an observer by your side, helping you look back at what you've experienced and why you felt the way you did. That's what I want to build: an AI companion that feels deeply personal and deeply private.
2.4 Keeping the original audio, and optimizing storage and battery use
Some products built around all-day recording focus on transcripts and summaries, without keeping the original audio long term. But just like the Apple Watch concert recordings I mentioned at the beginning, there are so many moments when you really want to listen again. A meal with your family after you've been away from home for a long time. The slightly nervous, slightly awkward conversations on your first date with the person who's now your partner. Reading a summary and hearing those voices again are completely different experiences.
Or think about graduation day. From meeting everyone in the morning to the end of the evening gathering, an entire day of laughter, well wishes, and goodbyes is something worth keeping. A year or two later, you might suddenly want to look back at that part of your life and feel the atmosphere of that day again. You could open that day's recording and actually hear it. That's a wish of my own, and I really want to make it possible with Auday. It's why I've insisted on keeping original audio playback: so that one day, you can return to a day in your past, with more than just an AI-generated summary left of it.
Of course, keeping audio over the long term means taking storage and battery optimization seriously. Auday is designed around the needs of recording throughout a 24-hour day, with optimizations for battery use, audio compression, and how silence is stored, to save as much space as possible while retaining the sound details you'll want when listening back. Based on an estimate for everyday use, 30 days of recordings take about 3.5 GB. To me, keeping the original audio and making the app practical to use long term have to go together. It shouldn't just be fun when you first install it, only to leave you worrying about battery and storage after a while.
(Continued in the next reply — part 4/4.)
Product Story — continued (4/4)
I've also added an audio retention window. You can choose how long to keep recordings—six months, for example. Older audio that has already been successfully transcribed can be cleared once it's past the window you've set, while transcripts, summaries, and other AI-generated text remain. That lets you find your own balance between listening back to original audio and saving storage space. The textual memories you build over time can still be searched and revisited. This was always on my mind while designing Auday: it has to be practical for long-term recording before it can gradually deliver the value I hope for.
3. Product thoughts and a few personal ramblings
This is my first time doing iOS development, and my first real step as an independent iOS developer. Over the past month, I've spent a huge amount of time iterating on this app, pretty intensely. If there are things I haven't done well, or anything that feels awkward when you use it, please tell me. The next version has already been submitted for App Store review and is currently expected to come out next week. It will keep improving the storage design to use less space than it does now, and optimize performance so the app feels smoother. At the same time, I'm running lots of experiments to see whether there's a better agent architecture and whether I can make better use of context. On privacy, I also want to keep exploring whether this design can go a step further.
This isn't just software that's finally available to download. It's an idea I've looked forward to for a long time, made real. It's also the first time I've followed a real need of my own this far: trying things, exploring, and turning it into a complete app. That feels pretty special. I've been excited the whole time I've been writing this, wanting to tell everyone about these discoveries. I also designed four widgets for Auday, because I love beautifully designed iOS apps and want mine to look beautiful too, haha. So lately I've been doing a lot of catching up on design. I hope that, as I keep iterating, Auday will get better not just in its features, but in its design and how it feels to use.
The website is https://auday.ai/, with more examples, screenshots, and more detailed introductions to different use cases. If you'd like to learn more, please take a look, or go straight to the App Store (https://apps.apple.com/us/app/auday-all-day-watch-recorder/id6814210373) to see the listing.
4. What I want to do next with Auday
Finally, a little about where I want Auday to go. Recently, from Muse and Dot to CUE, more and more people have been exploring personal agents. Our expectations of AI are also shifting from “a tool I use” toward “an individual that can do things for me.” We talk about AI assistants and AI secretaries because we want something more like a real assistant: one that understands what we're asking for and can take care of things on our behalf. I think that's a great direction. It can make us more productive and change the experience of using AI quite dramatically.
But as we rely on AI more and more, I keep feeling there's a missing place between the user and these personal agents—a place that's more private. If we think in terms of closeness and levels of trust, I want Auday to sit in the innermost layer, with the same level of privacy as our diaries. The agents outside can each have their strengths and help us get things done. The layer closest to us should know us well enough, be private enough, and feel safe enough to trust. It would know what matters to you and why you want to do something, then, when you choose to, pass the task and the necessary background to those outside assistants.
That's the broader system I want to explore next. Outward-facing tasks that need doing could be organized and delegated through Auday. Inward-facing, more private feedback and context could return here and become part of the memory that helps it understand you. I think this is the layer that's missing, and I want to build it. It might not be the most capable of all your agents, but it has to know you—as much of you as possible—while still giving you peace of mind. You shouldn't have to hand all of yourself over to every outside service just so AI can understand you.
Of course, setting that bigger vision aside for a moment, even the recording, memory, sharing, and import/export capabilities Auday has now open up plenty of possibilities to think about. AI-powered social experiences, digital versions of ourselves, or richer personal profiles built from long-term records of our lives, with new ways to understand ourselves along the lines of MBTI, and eventually connections to all kinds of productivity tools. An AI that stays with you over time and really knows you could grow in so many directions. At the beginning, I was just trying to find an AI recording device I was happy with. The more I build, the more possibilities I find worth exploring. I hope you'll join me in looking forward to what Auday becomes next.