Highlights:
- Why I finally chose Apple Watch
- Auday's features and my design rationale
- The solo creator model is still viable from zero
- What Auday plans to do next
1. Introduction
On the first day of the National Day holiday, I launched Auday on the App Store. It’s a software that turns your Apple Watch into an AI recorder, priced at a one-time purchase of $19.9 USD, with a 50% discount ($9.9) during the launch period.
This was my first time doing iOS development. Previously, I created two open-source projects that both hit #1 on GitHub Trending. Many people knew me as "BaiFu" back then, and some called me a "marketing master." I always felt a bit defiant about that label. So this time, I wanted to start completely from scratch. I didn't post on my personal account or return to my familiar promotional comfort zone. Instead, I registered a new X account as the official Auday handle, leaving behind my developer history to see if the product could stand out purely on its own merits and pain-point solutions.
When it came time to prepare the introduction post, I was admittedly nervous. A brand-new account with no prior engagement might easily get lost in the noise. I had mentally prepared for silence. But Auday was built piece by piece, reflecting my deep understanding of Apple Watch recording and addressing past user frustrations. I believed it had the potential to become a hit based solely on the product itself.
I also left myself an escape route: if nobody cared during those seven days, I would wait until the software passed local filing requirements and launched in China, then shamelessly promote it there.
On day one, I posted the first introduction on the new Auday account. It was formal, story-free, simply showcasing features and advantages. I spent $200 on ads. Then I handed the account over to Dot (an AI agent), instructing it to reply to comments and engage, and stopped looking—mostly because I was too scared to check. That was the entire launch effort.
https://x.com/auday_ai/status/2106281145545719983
Unexpectedly, within 24 hours, the post’s views skyrocketed past 1 million. Auday shot to #1 in paid apps across multiple countries on the App Store, becoming a viral sensation. Major influencers and even official accounts began reposting it. As of now, the official account has over 3 million views, and Auday generated roughly $300,000 in sales during the holiday week. If calculated as ARR, that’s tens of millions (though that’s not quite how you should calculate it, haha).
This time, people got to know Auday the product first. Starting from one intro post, recognition grew through shares and validation. Seeing the product recognized on its own merits is what makes me most proud.
However, by October 4th, seeing how well Auday was doing, I couldn’t resist posting about it on WeChat Moments. Screenshot below. This shouldn’t count as premature self-promotion, right? Haha.
Minutes before publishing this article, I couldn’t help but post again: Auday reached #1 overall in Japan. Exciting news, perfectly timed for October 10th—a date symbolizing perfection.
Screenshot from Oct 10
Today, I want to officially claim Auday in this article and introduce this new tool properly as its developer. Let’s start with the initial spark, why I chose Apple Watch, and the journey of exploration and thought along the way.
2. Tried Many AI Recording Devices, Finally Chose Apple Watch
2.1 Unsatisfied with Existing Hardware
Over the past year, I bought at least five different AI recording hardware devices, ranging from popular brands to niche, avant-garde ones. Unlike traditional voice recorders, these devices automatically summarize and use AI to organize recordings quickly. The category became popular because it solves a real need. For many, including myself, recording is about creating a safety net: you might not listen back immediately, but you want the data available when needed. However, if you record everything indiscriminately, you end up with hundreds of files you can’t find. The advantage of AI recording is organizing this chaos. For people who forget things easily, have short attention spans, or struggle with ADHD, this help is significant.
So, I’ve been exploring this space to find something truly suitable for long-term use. My first purchase was an AI recording card that sticks to the back of the phone. It was trendy at the time. But even with a thin case, it added bulk that made holding the phone uncomfortable. I later found an ultra-thin version, but combined with a phone case, it still felt awkward. Despite being very thin, that extra thickness mattered given how often we use our phones daily.
Then there was charging. I already charge my phone, iPad, and watch daily; adding another device to charge was inconvenient. Usage frequency dropped—I only attached it before important meetings. I tried recording beans and stylish accessory-like recorders, but they all faced the same issues: audio quality, portability, social awkwardness, and the psychological burden of carrying yet another device. Plus, the critical issue of privacy.
Despite these frustrations, I remained passionate about AI recording hardware. Current AI agents and harnesses can contextualize digital activities, but their connection to real life is weak. Real-world conversations go through humans before reaching screen-based AI. If AI could directly remember these interactions, I wouldn’t need to relay them constantly. The context AI sees and the tasks it can help with would expand significantly. This is why I remain hopeful about AI recording.
Companies in this space are pursuing similar goals: longer battery life, better audio, stronger AI. Some explore all-day recording or integrate powerful agents linked to screen-based AI. But there’s a contradiction: the more you record, the greater the privacy concerns. Occasional meeting recordings feel different from uploading your entire 24-hour life. With so many startups entering this space, despite various privacy solutions, handing over your whole day still feels risky. I never found a completely reassuring option.
2.2 Discovering Apple Watch’s Potential
One morning, while checking sleep quality on my Apple Watch, it hit me: besides the phone, isn’t this the hardware I wear from morning to night every day? I’d been using it for a while due to fitness habits, so wearing and charging it was as natural as using my phone. I was looking for a device that adds no burden, and the answer was on my wrist. Initially just a fleeting thought, it ignited when I saw social media posts about using Apple Watch to record concerts. Below are sanitized screenshots of those posts.
Concert recording share 1: Identity/location anonymized
Concert recording share 2: Identity/location anonymized
I realized many people already prefer Apple Watch for concert recordings due to its audio quality. "Immersive playback experience" was a common theme in shared posts. The idea of turning Apple Watch into an AI recording hardware became irresistible. Worn daily, good audio quality, no need to buy/carry extra gear—the conditions were perfect. Why not try?
Deep diving confirmed my belief: Apple Watch is currently the best hardware for this task. After years of iteration, its hardware, battery, sensors, and software integration are mature. Startups can’t match this accumulation quickly. Initial battery tests gave me confidence to build an all-day AI recording tool.
Crucially, Apple Watch does more than record. Heart rate, sleep, exercise, real-time stress levels, and 80+ workout types fit into this small device. Once, trying yoga with family, it instantly detected the activity. I was impressed by its detection capabilities. Often, we don’t fully utilize these features until we start developing around them, discovering how much data is already inside.
Adding location info means the watch’s sensors have the potential to capture real-life context, physical state, and "you" as context. This excited me. Originally aiming for an AI recording hardware, I now envisioned an AI recording hardware that captures life context. Your life, body, health, and exercise connect with sound; conversely, these states help AI better understand recordings.
Traditional AI recording converts audio to text, letting AI find key points. But physical state is a direct clue. If stress spikes, that conversation deserves attention—maybe a crucial meeting or call. If heart rate differs from normal, and AI sees you chatting with someone you like, moving from office to a nice restaurant, combining these details enriches its understanding of "what I’m experiencing," linking it to a beautiful date.
Exploring these possibilities thrilled me. Apple Watch doesn’t just record well; it connects real-life states to audio. Scattered information works together here, exceeding my original imagination for AI recording hardware.
Surprisingly, previously sticking a recording card to my phone aimed to capture call audio via vibration microphones. Results were often poor, relying on noise reduction/transcription. I feared Apple Watch couldn’t handle this. During testing, wearing the watch and holding the phone with the same hand for a call, playback revealed I could hear the call audio. The watch’s proximity to the phone became a recording advantage. What required a special card was achieved by the watch. No reason not to proceed.
2.3 Accelerating Development Post-Keynote
The idea struck in May, but reality delayed progress. In September, Apple held its keynote. While excited about new products, seeing Apple Watch Series 12 made me restless. Apple announced Audio Intelligence, including Live Rewind (planned for year-end), which transcribes the last 15 seconds of conversation. Heart rate and HRV detection improved significantly. I’d been researching how to use biometric data to supplement recording context. Seeing Apple push hardware capabilities further excited me but also tightened my nerves. Apple seemed to realize the same thing: Apple Watch is ideal for AI recording. Happy my May idea aligned with Apple’s vision, but anxious because my engineering progress was only ~10%. Post-keynote anxiety drove me: I must build it now.
Apple keynote promo copy
Fun fact: Apple’s tweet title asked, "Tonight, stay up late or wake up early?" Indeed, excitement plus anxiety kept me awake all night after the keynote.
From September 10th onwards, I worked non-stop for a week, followed by continuous development, testing, and refinement, leading to today’s Auday. As a heavy user of AI recording hardware, having discussed experiences with many others, my goal was to create a product I’m satisfied with, solving common pain points. Having covered the background, let’s discuss what Auday does and why it’s worth trying.
3. Auday Features and Design Rationale
Briefly, Auday turns Apple Watch into an all-day AI recording hardware. Core functions like recording, transcription, and search happen locally on-device. Advanced AI features use your configured API Key. It’s a low-cost one-time purchase, no subscription, no account registration, and no backend server. These choices stem from my past year using AI recording hardware.
3.1 Extreme Privacy Protection
As software intended to accompany you long-term, Auday records not just sound but connects location, physical state, exercise, and health info. This involves massive privacy, making protection paramount and central to trusting a small developer. Building cloud services to compete with big tech on security is hard to guarantee. Avoiding this, I chose on-device processing.
Apple Watch and paired iPhone transmit recordings via WatchConnectivity, bypassing iCloud and Auday servers. Transcription uses local models on iPhone; search uses local semantic embedding models. From recording to reading/searching/playback, everything happens within your devices. Even without cloud AI, basic long-term recording needs are met.
Ideally, I wanted an AI subscription business. But implementing it required building a user system, routing AI access, and passing private content through my service. This raised two concerns: user identity/account info and private content sent for AI analysis. Even if vendors promise immediate deletion, users worry. Understanding this fear myself, I removed the user system and content relay server entirely. Local privacy version is a permanent one-time buy; AI uses your own API keys. Connect to self-deployed AI or trusted mainstream providers directly from iPhone. Summaries, highlights, reflections, or chats send relevant text/context to your chosen provider, keeping raw audio on-device. This trade-off reduces revenue but is worth it for this product.
3.2 One-Time Purchase Model
The buyout model stems from privacy considerations. Also, Apple Watch is excellent hardware that shouldn’t be limited to habitual uses. Often mocked as a "beautiful waste," many let it gather dust or only check health data. Yet, it holds untapped potential discovered through years of Apple’s iteration. Since it’s ideal for AI recording, I want existing owners to access this capability with low barriers.
Regular price is $19.8, launch discount at $9.99 one-time, lasting a long time. Friends joked the AI subscription costs exceed earnings. My response: irrelevant. Discovering Apple Watch’s potential as superior AI recording hardware sparked excitement, driving me to share this discovery with everyone.
3.3 Comprehensive AI Integration
Auday’s AI focuses on three areas. First, hourly/daily transcript organization and automatic highlight discovery. Like earlier examples, it references not just text but connected physical states, activities, and locations, finding noteworthy content comprehensively. Manual marking on-watch allows tapping the screen to flag important moments, guiding AI focus. Continue working, then review AI-generated highlights and hourly notes later.
Second is daily summaries and reflections. As an all-day observer, Auday combines speech, physical state, and experiences to offer third-party perspectives, uncovering overlooked issues and suggesting improvements. I anticipate this aiding personal growth alongside life recording.
Third is Auday AI Chat. Not just another Q&A bot. With rich life context accumulating continuously, it can become intimate. Everyone uses AI assistants/agents, but I want Auday AI to be the closest, most private, and understanding one. Paired with Auday’s exclusive memory system, it extracts long-term personal traits (preferences, habits, relationships, ongoing concerns) from hourly/daily reviews, storing them locally as personal memory for future interactions. Understanding accumulates rather than resets daily. The longer you use it, the better Auday AI knows you.
I love using dating as an example. Ask Auday AI, "Do Emma and I really like each other?" It references far more than dinner conversation. Comparing relaxed states with her against tense work states provides clues. Beyond spoken words, post-date reactions or car ride murmurs matter. As a long-term companion, Auday places events from different times into one context.
Combined with Agent architecture and context recall, it extends timelines. Records showing intense pre-date workouts and repeated body mentions allow linking efforts to impressing her. When you ask, "Do I like her?", it analyzes unconnected experiences with you. This interactive insight is exciting.
Thus, Auday AI leverages recording but exceeds it. Sound, physical state, location, dialogue history, and accumulated personal memory form its understanding material. It acts as a close observer, helping you reflect on experiences and feelings. Goal: the closest, most private AI partner.
3.4 Raw Audio Preservation, Storage, and Battery Optimization
Some all-day recorders prioritize transcription/summaries, discarding raw audio. But like concert recordings, many scenarios warrant re-listening. Rehearing a family meal after travel, or nervous first-date chatter, feels different from reading summaries. Graduation day laughter/farewells hold sentimental value. Years later, wanting to revisit youth, playing back that day’s audio lets you truly hear it. This personal wish drives Auday’s commitment to preserving raw audio playback, allowing returns to past days beyond AI summaries.
Long-term storage requires optimization. Auday optimizes battery, compression, and silent storage for 24-hour recording. Estimated 30 days of recording ≈ 3.5 GB. Preserving audio and ensuring long-term usability must coexist; avoiding battery/space anxiety after initial novelty.
Additionally, audio retention windows are configurable (e.g., six months). Old successfully transcribed audio clears after limits, retaining transcripts/summaries/AI text. Users balance audio playback vs. storage, while long-term textual memory remains searchable/reviewable. Designed for sustained recording to unlock expected value.
4. Product Thinking
First iOS development, marking my step as an independent iOS developer. Intense iterations led me to consider how to make Auday truly solve problems while feeling natural. Feedback welcome. Concurrently, experimenting with Agent architectures, context utilization, and advancing privacy designs.
Though not a professional PM, Auday’s development highlighted product thinking in specific questions: usage feel, positioning alignment, and genuine need fulfillment. For instance, the watch recording interface initially mimicked traditional waveforms. For 24-hour recording, constant visual feedback felt like the device was "eagerly listening," causing discomfort. Adjusting to slower, restrained rhythms conveyed a quiet, listened-to feeling. Considering bystanders, obvious recording indicators cause unease. Auday’s logo underwent lengthy deliberation: four lines representing four signals, symbolizing integrated analysis/judgment merging into one. These seemingly minor trade-offs shaped my product understanding.
In crafting Auday to solve real problems, I experienced its birth from idea to polished promo video via AI editing, a factor in overseas virality. Sharing insights from my first public account article; interested readers can link there.
Excerpt from first article, https://mp.weixin.qq.com/s/UyYVjlBCvQRJI6B_MmZbsA
Launching Auday realizes a long-awaited idea. Starting from personal needs, exploring, and building software feels special. Writing this excites me, sharing processes/discoveries. Official site auday.ai offers cases/screenshots/scenario details. Visit site or search Auday on AppStore.
5. Auday’s Future Plans
Finally, future directions. Products like Grok bot, Muse, Dot, CUE enter Personal Agent spaces. Expectations shift from "using tools" to "owning entities acting for us." Developers market Agents like matchmakers, highlighting strengths/helpfulness. AI evolution seems cyclical: starting with chat, adding tools/arms, becoming armored "mechs," now condensing capabilities back into dialogue as intimate personal assistants. Discussions of AI secretaries/assistants aim for realistic helpers handling tasks. Positive direction improving efficiency/experience.
Yet, amidst reliance, a gap exists between users and Personal Agents regarding intimacy/trust. Ideally, Auday occupies the innermost layer, as private as diaries. Outer Agents excel at tasks; inner layers must deeply understand/trust users. Knowing importance/motivations. External tasks organized via Auday, sharing necessary context with outer assistants when willing. Internal private feedback/backgrounds return here, becoming memory. Exploring this missing layer. Capability may not be strongest, but knows you best, preventing surrendering full context to external services.
Setting aside grand visions, current voice/body records, memory, sharing/import/export enable future ideas: AI social, digital twins, richer profiles from long-term logs (MBTI-style self-discovery), connecting productivity tools. Enabling a long-term companion truly knowing you to play larger roles in life.
Auday APP homepage screenshot
Writing late at night, Auday generated today’s Reflection titled "Meteor Landing on World". Fitting metaphor: initial desire for satisfactory AI recording hardware became real product, reaching users, opening unforeseen possibilities. Coincidence adds romance to excitement. Join me anticipating Auday’s evolution.
