In this podcast episode, we sit down with Patrick, founder of the '100 Bad Ideas' studio behind the StressWatch app, to hear the story of how they earned App Store features in over 170 countries and built a user base of roughly 4 million in just two years. 😲 Drawing on his experience at Meizu, Frog Design, and FITURE, Patrick shares a practical methodology for spotting opportunities, reflects on the hard lessons learned along the way, and offers insight into how today's app factory model differs from the mobile internet era—and who the new players are.

1. Patrick's Introduction and the StressWatch Success Story 🥳

Host Ronghui opened the episode by introducing Patrick as the developer of StressWatch and a former designer at Meizu and Frog Design, noting that the conversation would cover his two years of development experience and his approach to identifying opportunities.

Patrick described himself as a designer by background, having worked at Meizu, Frog Design, and FITURE before founding the StressWatch app at his product studio '100 Bad Ideas.' StressWatch is an Apple Watch–based app offering stress monitoring, health management, and watch face customization. Notably, StressWatch launched two years ago in April (relative to the podcast recording date in 2025), making this episode something of a second-anniversary retrospective.

Since launch, StressWatch has attracted roughly 4 million global users and earned App Store features in over 170 countries. Particularly striking: between February and March of the previous year, it hit #1 on Xiaohongshu's trending search. Patrick recalled being genuinely shocked when a friend sent him a screenshot of that moment. 😮

When co-host Koji asked why it trended, Patrick explained it was because of a post from a user who said an alert fired on their Apple Watch during a blind date.

During the date, this person was feeling intensely stressed, and our app happened to alert them on their Apple Watch that their stress was elevated. The comments exploded — "what app is this?!" — and people started sharing their own stress-monitoring screenshots from dates, relating deeply to the experience.

Another memorable moment came from Taiwan. One day, StressWatch downloads there spiked sharply — all thanks to a heated discussion on Threads. A user had shared their experience with the app, and before long professional doctors and prominent influencers had joined the conversation. StressWatch climbed to #2 in overall app downloads in Taiwan, briefly surpassing ChatGPT, which was sitting at #3 at the time. Patrick laughed as he recalled screenshotting that ranking — a genuine milestone. 😄 Today, StressWatch continues to perform well in international markets including Japan, South Korea, and Germany.

2. The Origin Story of '100 Bad Ideas' Studio ✨

When Koji asked about the story behind the studio name '100 Bad Ideas,' Patrick said it came from magic — he's a magic enthusiast in addition to being a designer, and was president of the magic club during college.

The name came about while Patrick and his co-founder Alex were brainstorming business ideas and listing them in a document. Around that time, Patrick listened to a podcast featuring Blake Vogt, considered one of the world's most creative magicians.

Koji joked that he only knew David Copperfield, so Patrick explained that Blake Vogt is highly regarded in the US — he placed near the top on America's Got Talent and, in college, was interviewed for a magic consulting position on David Copperfield's team.

During that interview, when Copperfield's team asked Blake Vogt whether he could guarantee at least one good idea, he replied:

I can't guarantee a single good idea, but I can guarantee a constant supply of bad ones.

Patrick recalled being struck by that answer. He came to believe that the essence of all creative work is not prematurely dismissing new ideas, but finding the innovative and valuable kernel inside them. Keep pushing bad ideas forward, and good ideas will inevitably emerge. Allowing failure, he emphasized, is what matters most.

So at the top of their brainstorming document, Patrick wrote: "First, let's come up with 100 bad ideas." When it came time to name the studio, a friend spotted that phrase and suggested: "Why not just call it '100 Bad Ideas'? That's literally what you wrote at the top of the document." And so the name was born.

Koji praised it as perfectly capturing a core principle of brainstorming — never negate an idea — and noted that it's a name that invites better ideas to emerge. Ronghui added that it feels distinctly "American," in the way Americans use humor and lightness to deflate the seriousness of things while adding meaning. Patrick laughed and said the friend who suggested it was, in fact, American.

3. From Designer to Founder: Patrick's Career Journey 🚀

Ronghui asked how Patrick's experiences at Meizu, Frog Design, and FITURE shaped and inspired his work on StressWatch today.

Patrick said his career started even earlier — during college, he worked at a small advertising agency specializing in the fishing industry. When Koji asked why fishing of all things, Patrick explained that the industry is small and fragmented, with many small companies that need advertising, leaving just enough room for a small agency to survive.

After graduating, a defining feature of Patrick's career was that he never worked at a traditional internet company in the conventional sense. Meizu and FITURE were hardware-software hybrid companies; Frog Design was a design consultancy. Working at Meizu in 2016 and Frog Design in 2018, he recalled feeling like he'd missed the last train of the internet boom — watching colleagues move to fast-growing companies like ByteDance and Pinduoduo.

When Ronghui asked what he specifically did at Meizu, Patrick said his focus was primarily on visual design: the visual and interaction experience of the phone's operating system. Meizu, he noted, placed enormous emphasis on user experience and design.

At Frog Design, the work varied by client. Having focused on interface design at Meizu, Patrick wanted to broaden his perspective — to do more user research and participate in the early-stage process of building products from zero to one. He joined Frog after a colleague's departure left an opening. There, he worked extensively on 0-to-1 innovation projects for large corporations and startups, learning a great deal about how to use user research to uncover product opportunities and potential feature innovations.

When Ronghui asked about Frog Design's user research methodology, Patrick described the Design Thinking framework, widely used by design firms like IDEO. In brief: research leads to insights, insights fuel ideation, prototypes are tested, and results are validated or revised toward a viable MVP.

Ronghui noted it sounded like a typical B2C product development process, and Patrick agreed. Frog Design's clients were mainly large enterprises, sometimes outsourcing projects that were either too risky or beyond their team's capacity. They worked with everyone from major corporations like Tencent to Series B and C startups.

Asked what "high-risk projects" meant, Patrick said primarily product innovation projects. For large companies, product innovation is inherently risky, so they hire external design consultancies to supplement ideas — and if the project fails, the agency takes the blame, he added with a grin. 😉

When Koji asked what the most creative project was, Patrick mentioned a car-washing robot. The concept: a user parks their car at a designated spot, and the robot circles the vehicle and washes it. The robot was roughly the size of a refrigerator with multiple brushes. A prototype was built, and the client even used it to secure investment — though whether the product ever shipped, Patrick never found out.

Koji asked whether, working on the "vendor" side, Patrick ever produced designs he was genuinely proud of despite ultimately serving the client's agenda. Patrick acknowledged that in an agency relationship, the client defines the goals and the agency applies professional skill to achieve them. There were projects he found satisfying, like the car-washing robot — but he grew increasingly dissatisfied with the fact that it was always the client who posed the problem, never him.

That's why when he left Frog Design, he set himself only one criterion: join a startup. After exploring a few opportunities, he chose FITURE — a company building sports-oriented hardware that excited him. He believed startups would offer more room for genuine innovation.

FITURE made a smart-mirror fitness device: when off, it's an ordinary mirror; when on, it merges mirror and screen into a single interface. It used a camera and AI to recognize users' exercise form, help count reps, and correct posture.

Ronghui mentioned it sounded similar to the product Lululemon acquired, and Patrick confirmed that a US product called "Mirror" existed but lacked the camera-based recognition capability.

Patrick worked at FITURE's domestic team before spending a year in New York handling US operations. That experience — conducting North American user interviews about lifestyle and health and fitness — had a profound impact on him. Although he eventually left FITURE, those years laid the groundwork for founding '100 Bad Ideas,' and he emphasized that no matter how many products they build, the core focus will always remain in "Grand Health" and health & fitness.

4. Differences and Opportunities Between Eastern and Western Health Markets 🌏

Ronghui asked which insights from Patrick's New York user interviews had surprised him most.

Patrick said he found several unexpected things in the health and fitness space.

First, the age demographic of users was quite different from what he expected in China. In Korea (and among current StressWatch users), the dominant age range is 20–35, whereas in the US the range is much wider and he often received feedback from older users. FITURE had projected its target buyer to be 30–40, but actual purchasers were overwhelmingly 40–50.

Patrick connected this to national development stages. The US baby boomer generation grew up during a period of economic prosperity, developed strong demand for health and fitness, and has the financial power to spend on it. Korea, by contrast, has a different demographic structure, leading to a younger cohort that spends on health-related technology products.

Second, fitness demand in the US is highly segmented, and users are willing to pay for even very niche fitness needs. For example, where the domestic Chinese market has fitness bands selling for a few dozen yuan, the US has premium products like WHOOP aimed at serious athletes.

The WHOOP band has no screen — intentionally, to fit a larger battery for high-frequency sensing and longer battery life. It sells for $200–300 and requires a roughly $20 monthly subscription on top of that. In Korea it's hard to imagine paying $200–300 for hardware plus an additional $20 a month, but for passionate American athletes, $20–30 a month is not a burden. The Oura ring, increasingly talked about, follows a similar hardware-plus-software subscription model with a $6 monthly fee — thoroughly mainstream in the US. There's also "Hatch," a bedside lamp that mimics natural light to wake users in sync with the sunrise. These hardware-software subscription bundles are common in the US but relatively rare in China.

Ronghui observed that Americans take health data very seriously — hospitals routinely ask about exercise habits, and Apple Health data can be accessed in clinical settings. Patrick agreed this reflects both the high cost of US healthcare and cultural differences.

Asked how these observations influenced the choice to build in the health space, Patrick said the direction wasn't clear at first. When ChatGPT became popular two to three years ago, he also seriously considered the AI space.

Before StressWatch, he built an AI Bible website to help Christian users ask scripture-related questions and deepen their study. After exploring many ideas and prototypes, he ultimately decided to focus on an Apple Watch–based stress monitoring product.

He identified several important opportunity signals in this space.

First, as the COVID-19 pandemic wound down, global Apple Watch sales had been rising on the back of demand for blood oxygen monitoring.

Second, as mentioned in the context of WHOOP and Oura, HRV (Heart Rate Variability) monitoring had become a key selling point in the US market. HRV reflects much of the body's recovery state and psychological stress, and stress, Patrick believed, is an especially important topic in East Asian cultures.

Patrick saw the core product-market fit as follows: HRV was a concept well-accepted in the US, but the underlying ideas of stress and physical recovery also resonated deeply with East Asian culture — yet no one in the East Asian market had built a product for it.

Third, surveying the existing stress-monitoring software landscape revealed that relevant apps within the Apple ecosystem were extremely rare, and the few that existed had unremarkable, often "masculine" designs. Smart watches like Garmin and Coros offer stress monitoring, but most just display a number like "stress index: 85" with no interpretation.

Patrick believed stress should be communicated in an emotionally resonant way. He saw an opportunity to design something emotional and appealing to female users. The convergence of three factors — growing wearable penetration, a gap in the East Asian market, and the potential for emotionally intelligent design to add real user value — led to his final decision to build StressWatch.

When Koji asked how emotional design manifests concretely in the app, Patrick explained that the most prominent element on the home screen isn't a number but a "Stress Bubble" — an animated illustrated character that changes based on the user's state.

For example, when a user is relatively relaxed (high HRV, low resting heart rate), a green Stress Bubble character appears lounging in a tube, wearing sunglasses, enjoying the sunshine. When stress is high, a red bubble hides in the shade of a tree surrounded by wind and rain. The environment around the bubble changes like "weather" depending on stress state: the ideal scenario is a clear sunny sky; the worst is a storm; middle states include cloudy skies or light rain. Through this illustration, users can intuitively grasp their current state and feel seen.

The watch app works similarly — opening it shows a dynamic emoji rather than a pure number like "stress score: 95." Patrick believed a raw number is of limited meaning; communicating the state and pairing it with an expressive visual is what matters. He also added diverse watch faces to let users personalize their watch experience according to their taste.

5. Ideas That Never Made It 💡

Koji noted that StressWatch is clearly a successful product, and said listeners would be curious to know which of Patrick's '100 Bad Ideas' showed promise but never got built — potential inspiration or hints for other founders.

Patrick laughed and said that some of the ideas he had logged were since built by others.

When Koji jokingly asked for the remaining 99 "bad ideas," Patrick mentioned one already realized: SunAlly, a watch sun-care app. Two years ago at Apple's WWDC, Apple Watch added an "outdoor sunlight exposure tracking" feature, and Patrick had the idea to combine it with sun health, skincare, and tanning — but never actually built it at the time.

With Apple Watch Ultra gaining dive functionality, he also considered building an Apple Watch dive app. He even traveled to Mexico to get dive-certified. In the end, though, he couldn't pursue it — Apple's dive API isn't publicly available to developers and requires a special application process along with stringent dive-industry credentials (extensive industry experience or professional testing environments). Patrick jokingly suggested that anyone who can get access to that API should try building it — and feel free to collaborate with them. 😅 That remains one of the ideas they wanted to build but couldn't.

Koji said all of these sounded fascinating, and Patrick added that he had also envisioned several AI-related products. The AI Bible project mentioned earlier was actually built, but they couldn't sustain the maintenance. The reason: it had been interesting at the time, but since neither he nor his partner are Christian, they lacked the product insight needed to nurture it. He later discovered that a team from Romania or another small European country had built a similar product called "Bible Chat" and was running it quite successfully.

When Ronghui asked whether he was already in startup mode at that point, Patrick said it was a transitional period — there was a gap between leaving his previous company and officially founding the studio. He confirmed that all of this predated StressWatch, adding that he had pondered what to work on even on weekends before officially leaving.

At the time, Patrick and his co-founder agreed on one key premise: they were willing to try new things, work on projects they cared about, and accept that they might have no income for two years. So they decided to run an experiment for a defined period.

Koji asked in disbelief whether they really were okay with no income for two years, and Patrick confirmed it. It was their own judgment call. Personally, he said he doesn't have strong material desires, and in the worst case he could return to his family home, eliminate housing costs, and get by on minimal spending.

Koji expressed genuine surprise that they'd committed upfront to two years without income. Patrick acknowledged it was a "privileged choice" — he had no mortgage or other loans, was unmarried with no children, and could sustain a simple lifestyle.

Ronghui asked whether Patrick's career arc — spending so long on the vendor side, then moving to the client side at FITURE — reflected a deep desire to build his own products.

Patrick confirmed it, saying he'd always felt driven to expand beyond his current limits. At Meizu, he was doing UX design but started thinking about the earlier research and demand-identification phases. At FITURE, he focused on new markets and new domains to find innovation opportunities.

In the end, I was no longer satisfied with solving problems that others posed. I wanted to be the one who identifies the problems and validates whether they're worth solving. That's the experiment I wanted to run.

He also recalled a memorable interview from another podcast — a travel vlog creator named Shifan who produces unique content exploring unusual and quirky locations. Shifan shared his creative strategy: give yourself a six-month deadline, accept that there'll be no income during that period, write a list of 20 pieces of content, finish them within six months, and if it doesn't work out, go back to a regular job. Patrick thought this was a viable framework and gave himself two years instead. Looking back, he said the two-year timeline was about right.

Koji noted that the podcast recording was happening near the end of Patrick's self-imposed two-year deadline, and congratulated him on reaching the life stage he'd envisioned.

Ronghui asked about the highs and lows along the way — especially in the early days when multiple paths weren't panning out. Patrick said he had set the two-year deadline and it hadn't expired yet, so he was relatively calm, but he was happy to describe the emotional texture of the process.

He said he started with excitement but also some drifting. He'd written down 100 "bad ideas" but wasn't sure which to pursue. With ChatGPT just emerging, he spent a lot of time thinking about AI, generating many AI-related ideas — but after careful consideration and market research, concluded that AI wasn't the right fit for them at that time.

Personally, Patrick believed AI will become as universal as water or electricity, transforming every industry. But he saw three specific issues.

First, the AI space exhibits a strong winner-take-all (Matthew Effect) dynamic — ultimately one or two companies will dominate model development, making it hard to compete with OpenAI without deep pockets and top-tier talent. Not the right path for their team.

Second, if AI is infrastructure, user data is the core asset — and there's a fundamental paradox for startups who don't yet have user data.

Third, returning to what he knew well and finding connection points there would yield many more product opportunities.

But there's a paradox here. If I return to familiar territory, I might not need AI at all. If a domain is familiar and the opportunity exists, integrating AI might not actually matter.

So they ultimately built StressWatch. After launch, positive signals came in — responses to Xiaohongshu posts pre-release, user activity in the closed beta group — that confirmed the product had potential. Seeing users discuss the product in the group every day gave him real hope.

One PMF test approach: invite users to a group, release an MVP beta, and observe without prompting whether anyone is still talking about the product after one or two weeks. If no one discusses it, that's a true failure, he joked. 😉 But at the time, users were chatting about it daily — a major source of encouragement. User growth and Apple features followed soon after, providing further validation.

However, Patrick admitted he made the mistake of underestimating the product's potential. Mainly a failure of imagination — he assumed the product was limited to Apple Watch users, and since that was a finite audience, he expected limited overall growth. He thought 100,000–200,000 downloads would be a fine result.

Because of that low expectation, after launch he put minimal effort into feature development and operations beyond basic bug fixes. He wasn't completely inactive — he collaborated weekly with Xiaohongshu KOLs — but he never gave it full effort. His energy was scattered across other opportunities: other work projects, the dive app, and various experiments.

When Koji asked whether, looking back, going all-in on this one thing from the start would have produced significantly better results, Patrick said if he could go back, he would do two things differently.

First, channel all his R&D and design experience into making the product more refined and deep.

Second, raise capital — not necessarily VC, but from friends or other sources — to invest aggressively in growth.

Koji asked why he specifically emphasized "not VC," and Patrick explained that he believes VC favors high-uncertainty, high-return projects.

Our product is ultimately a utility-type, cash-flow product — not the kind of high-multiple return VCs expect. There are many other capital sources out there. Someone once mentioned "3F"…

Koji explained: Friends, Family, and Fools.

Patrick agreed, adding that distribution company capital is another option — many distribution companies invest specifically to facilitate product distribution — or you can self-fund. But more importantly, he said, was the need to have focused on this product more intensely at the time. The lack of focus created a roughly six-month gap. During that window, competitors noticed StressWatch, built similar products, or added the same features to existing apps — and in those six months, their products and stress-monitoring additions grew faster than StressWatch.

Ronghui said six months is a long time. Patrick agreed, but noted that fortunately the competition wasn't too fierce. If he could do it again, he said he would never waste those six months.

Koji asked whether the early lack of focus was partly about hedging risk, and Patrick said yes — he was also exploring new possibilities at the time.

But the deeper realization, he said, was that when a product has a clear PMF — no matter how small it seems — it's worth going all in. He simply hadn't understood that yet.

Koji asked what the PMF looked like quantitatively at that point — for example, what was the advertising ROI. Patrick said that in the early days, discussion was concentrated in specific pockets rather than spreading broadly.

Without active operations, daily revenue was barely a few dozen dollars. Combined with platform constraints around Apple Watch, he expected limited room for growth — perhaps enough to cover living expenses. Patrick acknowledged he had clearly underestimated the ceiling for wearable products, partly because there weren't many comparable success stories at the time.

Koji asked how large the biggest commercially successful Apple Watch apps in the space had gotten. Patrick cited Welltory as an example — founded years earlier, now with around 10 million users. While exact revenue figures aren't public, the global Apple Watch user base is clearly in the hundreds of millions. Theoretically, an Apple Watch app has the potential to reach tens of millions of users. That's nothing to a large company, but for a small team, tens of millions of users is already a substantial achievement.

Ronghui said Patrick's perspective would be a valuable reference for indie developers. Between the seemingly contradictory choices of "keep trying" and "go deep," what advice would he give to someone in a similar position?

Patrick said the first thing is to pay close attention to whether your target users have strong needs, and whether those needs are quickly validated after launch — through discussion volume, conversion rate, or similar signals.

No matter how small the need looks, if it's a strong need, it's worth investing time and effort. This aligns with the argument from the Innovator's Dilemma. The best starting point for a startup is a domain where large companies can't generate satisfying growth rates, but where a strong need clearly exists.

Find that domain, and it's worth going deep, he said. Even an Apple Watch app can grow to millions of users, and for a small team, that's already meaningful.

Ronghui asked what emotional or rational signals told him it was time to commit. Patrick named several emotional signals: discussion volume among users, growth rate, and the fact that early on many bloggers spontaneously promoted the product — for free or in exchange for simple resource swaps like promo codes. That kind of organic support was invaluable for a cash-strapped early team.

From a rational standpoint, he analyzed quantifiable data: whether Day 1 retention hit 30–40%, whether Day 7 retention was around 15%, whether retention data met threshold criteria, and how difficult new user acquisition was.

6. A Methodology for Spotting Product Opportunities 🔍

Ronghui mentioned that Patrick had previously shared a framework for finding product opportunities, and asked him to walk the audience through it in detail.

Patrick said that through building multiple products — with StressWatch and PeakWatch being the two successful ones — he had distilled a few key ideas.

First, for small teams, it's better to think in reverse from MPF (Market-Product Fit) rather than chasing PMF (Product-Market Fit). PMF is a great-sounding concept, but it assumes you start with a product and then find a market for it — appropriate for something like Notion.

Koji joked that people who find Notion too complex just haven't started using it yet — once you do, it feels natural. Patrick said his own documentation needs aren't strong and he's mainly design-focused, but acknowledged Notion is an innovative product that imposes a certain learning curve. Products like that are concept-forward, and usually require venture investment and a relatively long validation runway.

The reverse approach is to start by observing what already exists in the market. Copying a product from one country and launching it in another was common in early Chinese internet. Another approach is finding information asymmetries — like StressWatch, where Patrick identified an awareness gap around HRV between the US and East Asia, recognized that the existing apps in the space were mostly scoring a 60 out of 100, and believed that through design and holistic optimization he could push the product to 80–90, thereby entering an underserved space cleanly.

Many developers reference app store rankings when building. Patrick admitted they made a similar mistake — seeing a product in a top position that didn't seem particularly complex and thinking they could build a better version. The problem: making something 5–10 points better than a competitor is often not enough to make users abandon their ingrained habits. The cost of switching must be lower than the value the user gains.

As an extreme example, if your product is 10 points better but the competitor already has a large user base, when you promote your product those existing users will defend what they use — making distribution much harder. Patrick advised drawing inspiration from competitors but not copying them directly.

When scanning app rankings, I suggest looking at products ranked roughly 200–500 in a specific subcategory. That signals some user demand exists, but competition isn't fierce. If a product with mediocre quality is still sustaining users, there may be a real opportunity there. Top-ranked products have earned their position for good reasons — they're not easy to leapfrog.

Koji noted that focusing on mediocre products that are already inside the top 500 is a genuinely interesting lens. He asked whether there were other approaches or mental models Patrick used when hunting for the 100 bad ideas. Patrick added:

What he just described was about reading rankings. The second point is don't be afraid of small demand. He once came across a useful framework: the horizontal axis is population size, the vertical axis is demand intensity. Applying a "Pareto 80/20 rule" lens produces four quadrants.

• 80% of people with 80% demand (strong needs for the majority): Already dominated by large companies — ride-hailing, food delivery, messaging. The Meituans and WeChats of the world already own this space.
• 80% of people with 20% demand (weak needs for the majority): Suited for large companies' secondary businesses. They can use main-business traffic to lower acquisition costs for these sub-businesses.

When Koji asked for an example, Patrick cited Mobike shared bicycles — a universal but not particularly strong need, where users are willing to pay but only a limited amount, making it ideal as a sub-business under a Meituan umbrella.

For small teams, the best fit is the '20% of people with 80% demand' quadrant — a smaller user base with strong, clear needs. These users are few but their need is real. A passionate athlete's health monitoring need, for example, is very strong — the kind of person who goes to the gym two to four times a week and specifically chooses hotels with gym facilities when traveling.

Another compelling example is a food recognition app for diabetics. Ordinary people may not care much about identifying the composition of food from a photo, but for diabetics, knowing the precise nutritional breakdown of what they eat is a health necessity. A product like this finds exactly the right niche — small user base, strong and stable need, perfectly capable of sustaining a small company.

The last quadrant is '20% of people with 20% demand' — needs that don't suit a company's core product, or sometimes any company at all. But as AI raises programming efficiency and lowers development costs, these needs may gradually get served; for founders, though, they should be low priority.

Patrick added a reflection on AI specifically. He thinks AI is well suited to people who already have some product in market or easy access to a specific community. Post-AI, people tend to start from a technical vantage point — macro trend assessment, product ideation, then specific users. But the reverse — starting from a deep understanding of a specific user group's characteristics and needs, then designing the product accordingly — lowers acquisition costs and produces products that genuinely fit real needs. It's often more effective.

For example, after StressWatch, they launched PeakWatch for Apple Watch fitness enthusiasts. Because they already had a StressWatch user base, the new product had no "cold start" problem — they could test directly with existing users.

The book Small Giants describes a similar approach: find a nearby community (neighbors, a club), observe their needs, build a product or service for that community, and scale gradually from there.

Finally, Patrick stressed that staying attentive to new technology trends is critically important — especially for iOS developers, for whom Apple's annual WWDC is essential viewing. The reason StressWatch earned global App Store features was that Patrick studied the WWDC announcements repeatedly and built around a new mood-logging API Apple had introduced. The API let developers more easily log emotional tracking data to HealthKit; Patrick built a mood-logging feature in StressWatch using it, aligned precisely with Apple's promotional direction at the time, and earned worldwide editorial features as a result. The sunlight exposure feature followed a similar path. And when Apple introduced desktop widgets, theme-customization companies surged overnight.

Koji noted how important it is to follow Apple's footsteps, and Patrick laughed that following AI's footsteps matters too — though AI can sometimes backfire. He'd recently seen a meme about OpenAI threatening two-year-old startups.

7. Learning from Mistakes: Lessons from the Founder's Journey 📉

Koji asked which of Patrick's many frameworks came from hard-won lessons rather than theory. Patrick said nearly all of them came from difficult experiences.

First: when you find PMF, don't hesitate — go all in. Second: small teams don't need to fear small needs. In large internet companies, projects need to target massive markets, but for a small team, serving even tens of thousands of paying users well can fully sustain operations.

Product experience isn't the deciding factor — what matters is finding a niche where the team's strengths can shine, not bolting features onto already-mature products. He also stressed that growth strategy is just as important as the product itself. Patrick and his co-founder came from hardware-software and consulting backgrounds and lacked experience in pure software growth — a gap they had to constantly close. Getting from 0 to 1 with a product is just the first step; the journey from 1 to 100 may require hundreds or thousands of times more effort and resources.

Koji asked whether Patrick considers growth opportunities and product opportunities equally important, and asked for a growth opportunity he identified early that paid off. Patrick said StressWatch was a good example. When he noticed strong domestic traction and a gap in East Asian markets, he moved quickly to go international. In the early days, lacking overseas experience, he simply started buying ads internationally.

Every product has a window period — not just for users, but for product growth too. In the early days, acquisition costs across many countries were extremely low because there was no similar competition. Now those costs in those markets have risen two to three times.

On the international expansion front: while China and the US are large markets, targeting smaller markets first — Japan, South Korea, or European countries — can be a good strategy, as those regions have strong purchasing power.

Ronghui asked what mistakes he made on the growth side. Patrick listed two.

One: not going all in on growth early enough. Two: insufficient understanding of how to grow an internationally-facing product, and inadequate resource investment. It wasn't until over a year post-launch (around June of last year) that he began scaling ad budgets across multiple countries. In hindsight, that was far too late. Their current strategy: as soon as a product has validated PMF and has no critical bugs, immediately launch that international growth effort.

Ronghui asked about Patrick's plans for StressWatch going forward. With a new product also in development, would he focus exclusively on StressWatch and that new product, or also build more?

Patrick said they currently have three products: StressWatch, PeakWatch, and LianJiu (练就), and would be spending significant time on all three this year. The team is split into two groups — one for StressWatch, one for PeakWatch — with about 10 people total (full-time and part-time).

Future direction will depend on where the product and company are in their lifecycle. For example, utility health products reach a point where adding too many features is counterproductive — especially for international users who prefer a product that solves one problem well. So options might include investing energy in new products, co-developing with other teams, or acquiring small products or teams.

Ronghui asked whether Patrick was open to being acquired, and he said not for now. But what matters most, he said, is people.

I believe the success of a product and an organization comes down to who the people on the team are.

Patrick said all of his current energy is going into the existing products, and building new ones requires finding the right new people. They're open to various forms of collaboration — product partnerships, corporate cooperation, acquisitions — and interested parties can reach out through the podcast team.

8. Global App Factory Models and What They Tell Us 🏭

Koji observed that once you have one successful product, the natural instinct is to build a second, a third, and gradually scale into an app factory. Many such factories already exist, domestically and internationally. He asked whether there were any that Patrick particularly respected — or aspired to become — and whether the word "factory," with its connotations of mass production and soulless copy-paste, could coexist with genuine quality.

Patrick said the app factory world is vast, and that while they're still a small workshop, they're aiming to gradually build toward a factory model. The specifics of how to operate it are still being worked out.

In terms of mature app factories he's observed, Patrick pointed to AIBY as one example — an early mover with strong product operations and impressive monetization capabilities. There are similar companies in Vietnam and Turkey, and some have actually cloned StressWatch. Many of these companies, he noted, are excellent. The word "factory" might imply low-quality output, but if you download and use their apps, every product is highly polished.

Starting with a domestic example: he introduced a Beijing company called Enerjoy. Not much public information exists about them — these companies tend to avoid the spotlight and operate quietly without raising external investment. Enerjoy has shipped multiple products, and using them firsthand reveals extremely high overall design quality, sophisticated animations, and sharp monetization strategies that guide users toward paying.

For example, Enerjoy's MePlus app is, at its core, a to-do list app. What they noticed: traditional to-do apps are designed for efficiency-focused office workers. But many homemakers also need task management — and almost no product was designed specifically for them.

Beyond pure productivity, people have strong emotional and self-improvement needs as well. The App Store screenshots for this product apparently feature a homemaker doing housework.

Ronghui laughed and said homemakers are probably busier than office workers. Patrick confirmed the app is essentially a to-do list for homemakers and has an enormous download count.

Koji asked what distinguishes it from a regular to-do list. Patrick said: first, the design is extremely cute and fun. Second, it cleverly bundles diverse features.

Beyond the basic to-do list, it offers many pre-built template lists like "personal growth list" or "5 habits that transform your morning." These are ideal for users with self-improvement or lifestyle goals, lowering the barrier to engagement. Patrick said the app's paid conversion rate is exceptional — that after reading the in-app copy, you'll feel a powerful urge to think, "I have to subscribe to this."

Ronghui said it reminded her of the calendar and countdown apps made by a well-known domestic company during the peak mobile internet era. Koji specifically mentioned the "Days Matter" countdown app.

Patrick spoke about the first-generation app factories — a time when the App Store had just launched and user demand was exploding. Those teams produced many excellent apps, earned official Apple features, and benefited from massive organic download demand. A defining characteristic of that generation: users were acquired almost entirely through organic discovery. Over time, however, the total number of apps in the market kept rising and organic acquisition declined — setting the stage for the emergence of second-generation app factories.

The rise of the second-generation was closely tied to the global spread of overseas distribution infrastructure. Platforms like Facebook and TikTok built global audiences and became massive advertising platforms. Companies — especially early movers — could reach global users at relatively low acquisition costs through Facebook and TikTok. That was a powerful infrastructure advantage.

Domestic examples include Enerjoy and also Rui Qi (睿祺), which built the plant-identification software Xing Se Shi Hua (形色识花) and recently generated a lot of discussion. Thanks to advertising platforms like Facebook, these companies could reach global users, particularly those with high willingness to pay and strong monetization potential.

Koji asked about the difference between first- and second-generation app factories, and why the first generation didn't evolve. Was it purely a growth capability gap, not a product capability issue? Patrick answered that growth capability is genuinely very hard to build. Looking at apps similar to MePlus: MePlus wasn't actually the first product of that type. Before it, there was already a self-improvement-themed to-do app with beautiful illustrations, which received many Apple features and grew primarily through organic discovery. But they failed to adapt to the shift toward ad-driven growth later on — and over time, their user base shrank and their influence faded.

Koji noted that Shizi Lu (十字路口) podcast was about to publish an article with "10 things you didn't know about Rui Qi" — a company that maintained a high profile of mystery for years before recent revenue disclosures sparked widespread discussion across platforms.

Patrick said, having covered the domestic side, he'd now look at overseas companies. Internationally, he has high regard for Lumi, a French company. As a designer, he has a natural affinity for products with exceptional design, and Lumi's website and all of their app products demonstrate very high design quality. They focus primarily on fitness and health products.

Another company worth mentioning is Genesis, a Ukrainian studio with many products, including BetterMe — consistently ranking well in health and fitness charts across many countries.

He also mentioned Voodoo — technically a gaming company, but with an operating model Patrick finds fascinating. Voodoo is a French hypercasual game studio that both develops games and offers publishing services. When Patrick connected with them, he found they have extremely strong growth capabilities — strong enough to offer publishing services to external developers.

Their internal innovation mechanism is also distinctive. According to one profile of the company, their internal product division is split into two teams: one focused on prototyping, and one responsible for running finished products. The prototype team is made up of the most creatively driven talent obsessed with gameplay innovation. They continuously experiment with novel game mechanics, and once something shows promise, it's handed to the growth team for early ad and growth experiments. If a gameplay mechanic's test data looks good, it immediately moves to the commercial team, which develops it further, designs monetization, and scales it. Patrick said he finds the separation of prototype and product-commercial functions fascinating — and likely a well-suited management model for their game business.

Ronghui compared it to an incubator, and Patrick agreed. He also mentioned Abishkking, a Chinese company performing exceptionally well in the fitness app space by using a "reskin" strategy — building variations like workouts for women's abs, women's chest muscles, men's abs, men's chest muscles, and so on. During the pandemic, they occupied half of the global top 10 health and fitness app download charts.

Koji said the strategy sounded wildly imaginative and somewhat audacious. Patrick agreed, and said that while all of these companies are called "app factories," the strategies and models they adopt are very different.

Some companies compete on exceptional product quality and sharp market insight; others excel at self-replication to build comprehensive product portfolios; and some — like Abishkking — compete primarily on commercial monetization capability and extremely high conversion rates. If you're struggling to improve paid conversion in your own product, Patrick advised looking at Abishkking as a reference.

Koji said Patrick had shared an extraordinary amount of information today, and that digesting it all would take days. Combined with the opportunity-spotting methodology discussed earlier, it would offer genuinely rich insight and inspiration to everyone listening.

Ronghui closed with a final question: looking back on the past two years, compared to peers who started around the same time, does Patrick feel lucky? He said yes, extremely lucky. Many projects that started around the same time as his have mostly shut down.

Cases like theirs — where someone not only built a successful product with growing users, but also maintained a team of nearly 10 people and became self-sustaining — are genuinely rare. Currently they're in a relatively stable position; competition exists, but it's not existential pressure. So yes, he feels very lucky.

On a personal level, he's experienced tremendous growth over these two years. The biggest shift, as he mentioned at the start, is that he's learned to pose his own problems, search for the answers himself, and develop independent value judgments about those problems. Second, his previous roles were concentrated in product design, leaving him without the experience to run a business holistically — but now he's fully involved in everything: hiring, project planning, studying how to grow a product, planning marketing budgets. A complete business operating loop has formed.

To put it most directly — and somewhat bluntly, he said — the central question is: how much do you invest each month, and how much revenue can you generate from it. As a bootstrapped team, they've always maintained a careful attitude toward spending. Whether it's hiring or buying ads, they confirm ROI meets expectations before proceeding to the next step.

Last June in particular, when they began pushing aggressively into international markets, the daily spend on ad accounts escalated rapidly and visibly. During that period, he went about a month barely sleeping. Every morning, the first thing he did was check whether the money spent had achieved the expected conversion.

Koji asked if that was because of the anxiety, and Patrick said mainly it was inexperience and the weight of investing real money. In the early days he had no experience with performance advertising and leaned on a knowledgeable friend for help. But it was his own saved money going in, and it was uncertain whether the investment would pay off. When a few days of ad spend equaled several months of revenue, the inexperience alone was enough to make anyone tense. After a month of operation with no losses — and with confirmation that the invested capital could be recovered within two months — he gained the confidence to keep going.

In practice, he noted, app factory-type businesses look good on paper but most of their capital flows through advertising platforms — making them fundamentally cash flow businesses. Unlike many internet products where network effects gradually lower acquisition costs, the products they're building are closer to utilities. As scale increases, acquisition costs may actually rise — because on advertising platforms, buying more volume tends to push CPI (cost per install) higher.

Koji noted that homogeneous competition drives up prices across the industry. He asked whether the subscription model's recurring payment nature gives these businesses high flexibility — specifically, whether a company that stops aggressive acquisition and adopts a conservative strategy can still rely on the existing user base's habits and spending inertia to generate a long, stable harvest period.

Patrick said some churn is inevitable — many users will cancel subscriptions. It really depends on the product's characteristics and its renewal rate performance. Some products attract subscriptions through impulse buying, but may not sustain renewal rates at an ideal level.

From a growth strategy perspective, companies generally keep investing in acquisition, but the key is judging when you hit an inflection point — when the cost of acquiring new users outweighs the value lost from churned ones. At that point, it may make sense to dial acquisition spend down to a minimum, allowing the product to settle at its naturally sustainable user size. That natural ceiling might, in fact, represent the product's true market boundary.

Ronghui asked whether Patrick thinks it's a good time to start a company right now. He gave a positive answer for three reasons.

First: the completion of global distribution infrastructure offers founders unprecedented opportunity. Looking at many app factory-type companies — including developers from relatively underdeveloped regions like Vietnam and Turkey — teams of a few dozen people have been able to build factory models within a few years, running dozens of products simultaneously. This is only possible because building globally has become remarkably accessible. Even markets like Africa, where you may never set foot in person, can now be reached effectively through modern infrastructure.

Second: AI-related capabilities have dramatically improved globalization efficiency. Multi-language support and translation were once weak spots, but AI now makes it possible to rapidly internationalize a product — whether that means shipping multilingual features or handling feedback and communication from users worldwide. AI can materially increase the efficiency of product operations and growth.

Third: with AI support, founders can iterate through product experiments at higher speed, dramatically increasing the efficiency of trial and error.

This point is especially important. Ideas themselves aren't scarce — everyone has many ideas that look good but may prove unworkable in practice.

An organization's core competency lies in how efficiently it can experiment and validate those ideas. AI raises that efficiency, strengthening the trial-and-error capability — and the more experiments you can run, the higher the probability of finding a product with real market fit.

Koji quipped that in the past it was the "100 Bad Ideas studio," but now thanks to AI, it could upgrade to a "1,000 Bad Ideas" or even "1,000,000 Bad Ideas" studio — because the cost of experimentation has dropped so dramatically. Patrick agreed.

Koji thanked Patrick for the great stories, methodologies, and all the useful, knowledge-expanding insights he'd shared, calling the conversation deeply valuable, and invited him back to the Shizi Lu podcast. Ronghui also expressed her thanks, and Patrick said he hoped to have another chance to come back.

Closing Thoughts

Patrick's StressWatch success story wasn't a stroke of random luck — it was the product of relentless thinking, hard lessons learned through trial and error, and a constant engagement with new technology trends. What stood out most was the spirit behind the name "100 Bad Ideas" itself: the refusal to fear failure and the drive to keep testing ideas — and, once product-market fit is found, the courage to invest boldly and move quickly toward global markets. It was a valuable conversation that gave fresh meaning to the ideas of "small needs" and "fast iteration" in the age of AI.