Stringers
Mobile App · iOS · written 11-6
피클: 나만의 AI 에이전트(Pickle) · Pickle Company

Americans have Muse, Koreans can't use it: an AI secretary that makes calls for you, hit #3 on the Korean iPhone overall chart in three weeks

What it is A Korean version of an AI secretary app. You message it what you need done, and it calls for you, orders and books on Korean websites, and schedules with your friends' AI.

Why it took off The US hype around Muse and dots made 'AI doing tasks for you' a hot topic, but Koreans couldn't use those (Muse isn't available in Korea, and US agents can't get past Korean websites). It filled that gap, and the founder's posts and invite codes on Threads did the rest.

BHow you can follow Find a language market where big US AI agents haven't launched or can't get past local websites, first nail the tasks locals hate most (calls, rescheduling, refunds, grabbing tickets), then copy the playbook: founder posts personally, closed beta, invite codes.

Grade B: the formula is clear, it's been proven in Korea, and it barely spent on acquisition. But Korea is now taken, Muse ranks #1 in Japan, and the rest of the markets you have to find yourself. Technical barriers and compute costs are high, and the window closes once big players arrive. The usable time is probably measured in months.
Data (updated daily) · Launched 2026-08-10 · Korean rating 67 · 11-08 · Best free overall chart rank #3 (Korea) · Charted in 11-08 stores 2
iOS · Overall Top Free rank · by store
102010-2810-3111-0311-0511-083
Data 2026-10-28 → 2026-11-08 (12 days, 12 observations). Lower rank is higher; gaps mean it wasn't charted that day
iOS · Number of stores charted (top 200 on any chart)
00.5110-2810-3111-0311-0511-081
Data 2026-10-28 → 2026-11-08 (12 days, 12 observations)
iOS · rating count · Korea
20406010-2810-3111-0311-0511-0867
Data 2026-10-28 → 2026-11-08 (12 days, 12 observations). Apple doesn't publish download counts; rating growth is a proxy for downloads.

A Korean founding team in the US rebuilt its personal AI agent, made in America, as a Korean version: it calls for you, orders on Coupang, books KTX, and schedules with your friends' AI. It spent no money on ads. The founder posted on Threads and handed out invites, and within three weeks it hit #3 on the Korean iPhone free overall chart. Why it took off, where the money comes from, what users are complaining about, and whether it can be repeated in another country are covered in the sections below.

A 39,999 won charge

On Nov 1, Pickle's founder Park Chae-geun (Threads account @danifestoo, about 6,000 followers) posted a thread in the voice of an ordinary user who had just installed the app: a few days earlier he'd paid 2,000 won for a personality test, then at midnight was charged another 39,999 won. A recurring subscription had been hidden below the payment page. He emailed for a refund, and the merchant replied: 'The terms clearly state no refunds are allowed, we can only help you cancel next month's renewal ^^'. So he told Pickle's secretary 'Luna': 'I was wrongly charged, check my email and get the money back for me.' Two hours later, the full amount was refunded.

The thread got 83 likes and 67 replies (Threads, 11-06 scraped). It isn't the most viral post, but it clearly shows what the product sells: the thing you don't want to do, can't do, or can't be bothered to fight over, you hand off to it.

App Store reviews say the same. One five-star review from Oct 18: 'It called a restaurant to make a reservation, and the owner sounded startled. The call transcript shows up live in the chat.' Another the same day: 'Grok Bot couldn't reschedule using aitocall or tello, but Pickle did it in one go. Moved.' Oct 28: 'A flight that could only be changed by phone, and it changed it for me.'

What it is

On the App Store as '피클: 나만의 AI 에이전트' (Pickle: My Own AI Agent), developer Pickle Company, YC 2025 Winter batch, available on iOS and Android, free to download. You tell it what to do, and it does it. For sensitive steps like payments and bookings, it stops and waits for your approval. The listed capabilities are:

  • Calls for you: changing phone plans, asking customer service, booking spots that only take phone reservations, requesting refunds. Calls are transcribed live, and you get a summary when they end;
  • Shopping and ordering: finding products, ordering and paying on Coupang and Daiso's mall, places you already use;
  • Bookings: restaurants, hospitals, lodging, flights, KTX. If tickets sell out, it watches for cancellations;
  • Dinner plans with friends' Pickle: both sides' AIs match free time and book the restaurant along the way;
  • Connects to services: Naver, KakaoTalk, Gmail, Google Calendar, government 24, carriers, food delivery, insurance, airlines.

According to Korean media tests, it connects to 60 many domestic and foreign platforms, including the National Tax Service's Hometax, Korail and SRT, and CatchTable (Digital Today, 11-04).

The company has pivoted several times: an AI stand-in for Zoom meetings, a desktop assistant, and an AR glasses product called 'Pickle 1', which US media questioned as vaporware (SF Standard, 2026-01-13). The agent app is its latest form. The US version launched on X on Oct 1, and the Korean version opened closed beta on Oct 14.

How fast it grew

MetricMetricSources
Korea iPhone free overall chart10-28 #23 → 10-30 #5 → 11-06 #3App Store charts
Korea iPhone Productivity free chart10-28 #4 → 11-06 #1Same as above
Google Play installs10-29 about 800 → 10-31 about 7,400 → 11-06 about 30kGoogle Play,11-06
Google Play Korea Productivity grossing chart#94 → #73Google Play charts
What the company says'Closed beta 3 weeks, 100k active users'Founder on Threads, 11-05,633 likes; self-reported, not verified

One thing doesn't add up: it hit #3 on the overall chart, yet has only 56 ratings on the Korean App Store. Our guess is two reasons: Korean users rarely leave ratings, and because it's closed beta, many people who installed it couldn't get an invite and never got in, so they didn't rate it. Google Play installs rose about 30k in a week, which matches the ranking. The ranking climbed gradually day by day, and several media outlets ran long hands-on tests during the same period, with no sign of inflated numbers.

还要说一句,团队很小。创始人 11-05 发招聘帖说:现在是「3 个高密度的工程师」在做代理框架、安全、基础设施和全部产品体验。

Why now: the US made demand hot, Koreans had nothing

In 2026 9, October, the US saw a string of launches: Meta's Muse (Sep 8), OpenAI's dots (late September), and Anthropic's Claude Duet (Oct 21). 'AI doing tasks for you' became a hot topic, and Korean tech circles talked about it daily. But Koreans couldn't use them: Muse wasn't listed on the Korean App Store, and even where installable, US agents couldn't get past Korean websites, which require security plugins, PASS mobile authentication, and a range of local payment methods.

The founder's Oct 14 launch post spelled it out: 'Korea has security programs, public certificates, and various Pay systems, unlike the US web environment, so we rebuilt it.' Then a post on the difference from Muse: 'We have Korean agent IPs and Korean connectors... Right now, something like Muse gets blocked from a lot of Korean sites.' Steps that must be done in person, like PASS authentication and payment, are handed back to the user.

In other words, the money in the education market is being spent by big US players, and it's the only one that can actually get things done on Korean websites. A reporter's hands-on conclusion: 'Domestic services that other AI tools can't get into, Pickle can handle.' (Game View, 11-06)

Where users came from: one post, 998 replies

We couldn't find its ads on the public ad transparency pages of Meta, TikTok, or Google. Not finding them doesn't mean it didn't spend, but the available signs point to free channels:

  1. The founder's launch post. The last line of the Oct 14 post: 'If you want to try it, just ask me for an invite code in the comments or DMs. Beta users get a big pile of free credits.' The post got 564 likes and 998 replies, most of them asking for invites. Every 'send me an invite' is free exposure for it.
  2. Invite links. The beta requires signing up through a pickle.com/i/… invite link, and each person can only invite a limited number. Someone already wrote a guide on 'how to get a Pickle invite' (foxcg) on Oct 29. The title of the first App Store review (10-14) is 'I really want an invite ㅜㅜ'.
  3. Steady updates and story posts. 10-23 posted 'The voice sounds more natural, daily call quota raised from 10 to 20 calls, usage cap ×1.5' (225 likes / 85 replies); 10-27 posted a privacy explainer (307 likes / 80 replies); 11-01 posted the refund story. Even a one-star review (10-29) ends up proving the channel: 'I spent a whole day scrolling Threads before installing it, and free GPT would've been better.'
  4. Friends bringing friends. 'Mom and Dad and my friends all got it installed, satisfaction 10000%' (10-29); 'Recommending it to people around me' (11-01). The feature for dinner plans with a friend's Pickle requires the other person to install it too.
  5. Hands-on media tests. Digital Daily (11-03), Digital Today (11-04), and Game View (11-06) each ran long hands-on tests, all after it hit the charts. They amplify it; they didn't start it.

Searching '피클 AI' on TikTok returns nothing but pickled cucumber memes. This product didn't take off on TikTok.

Where the money comes from: pay per use, not per month

Free quota resets daily at 4 a.m. Run out and you can buy a "reset ticket" that restores half your quota for ₩2,900, or all of it for ₩5,800 (about $2 and $4, 디지털데일리, 11-03). In-app purchases on Google Play run from $1.99 to $200.

People are already paying: it's ranked #73 on the Google Play Korea Productivity grossing chart and still climbing. Reviews include 'Willing to keep buying tickets to use it' (10-29). But iOS hasn't charted on any grossing list, so we estimate revenue is still very small (inference). Costs are not small either: every task requires the model to open a browser, click through pages, and make calls, and compute is paid for in real terms. Many reviews complain about the quota, like 'Day one was plenty, but on day two the quota drained fast before I'd finished anything' (10-24), which suggests it's controlling costs.

Who it's competing with

CompetitorsWhere it stands nowCompared with Pickle
Muse(Meta)US free overall chart #2, Japan #1 (11-06)The strongest rival, but not listed in Korea; media say the US is 'gradually opening up'
dots (OpenAI), Claude Duet (Anthropic)US, big-tech featuresWork-leaning: research, coding, long-running projects; no Korean local services
Grok Bot(X)#11,271 in Korea's Productivity chart (rating)Some users say it failed to reschedule by phone; Pickle succeeded
Manus#22 in Korea's Productivity chartGeneral-purpose agent, no phone calls, no Korean services
모두의 AI (Korean government + SKT, KT, Kakao)Closed beta in October, full launch by year-end, target 15M usersNational project, free; the beta leads with a chatbot, with action-taking agents coming later
ChatGPT, Claude, GeminiTop 3 in Korea's Productivity chartCan chat, but won't call on your behalf or place orders on Korean sites

Until early November, no second product in Korea can use Korean, complete tasks on Korean websites and by phone. Its moat is this time gap: if Muse enters the Korean store, or the full version of "모두의 AI" ships agent features in December, that gap is gone.

What users are complaining about

Lots of five-star reviews, but the complaints are specific, and they show how many of the people who came for the chart stick around:

  • Slow: "Logging into Coupang took 5 minutes and it still said failed" (11-03, three stars); "So slow, I'd rather do it myself" (10-22, one star).
  • Some connections fail: "CatchTable won't log in, it says it can't. Don't list it as connectable then" (10-22).
  • Calls flagged as spam: "It calls from a 02 number with no heads-up, my friends don't pick up, they think it's a scam" (10-24); "Pickle's number got marked as spam and people won't answer" (10-30).
  • Not enough quota, can't see how much is left: several reviews ask for a remaining-balance display.
  • Want more integrations: "Please add Toss, life would be so much easier" (11-04).

Most of these are engineering problems, not demand problems. Nobody says "I don't need this"; they say "if it were faster and more reliable."

If it were you

The takeaway: the formula to copy is this: a US big tech company hypes a category of AI product but can't get into a given language market. You go to that market and build a version that actually connects to local websites, identity checks, payments and phone calls. Pickle has taken Korea, and Muse is #1 in Japan, so avoid both.

Who it suits: a small team that can do browser automation and voice outbound calls, ideally made up of people from the target country who know which hurdles local sites impose (real-name verification, local payments, security plugins) and who have an audience on Threads, X or local communities.

Who it doesn't suit: anyone with just a chat wrapper; anyone who can't absorb the real compute cost of each task; anyone not in the target country who can't handle local phone numbers and payments.

How to do it: don't start with "does everything." Pick two or three things locals hate most that require a phone call or a local website, and do them well: rescheduling and booking, getting refunds, fighting with carriers and customer service, grabbing train tickets. Phone calls are the best place to start: the person on the other end is startled, and that makes the best demo footage. Charge per task, give a free daily quota, sell cheap top-up credits, and skip the monthly subscription. Handle spam flags on outbound numbers from day one.

How to get users: copy Pickle. The founder posted from their own account, ran a closed beta, and had people ask for invite codes in the comments, with a cap on invites per person. Then they posted updates every few days and a real story about "it got my money back for me." No paid acquisition at any point.

Timing: you need to be in before big tech opens the same product in your country. Muse is "rolling out gradually" now, so your window is probably measured in months, not years (inferred).

Signs you misread it: Pickle drops out of Korea's overall top 50 a month after its surge, and the grossing chart stays past #70, meaning people came for novelty and nobody wants to keep paying. Or Muse enters the Korean store and pushes it down, meaning localization only buys a time gap and isn't a moat.

iOS store page ↗

Research notes: competitor table, acquisition evidence, full research report
Competitors
NameWhereLaunchedScaleHow it differs from this one
Muse from MetaiOS2026-09-08US Free overall #2, rating 460k; Japan #1; not listed in Korea (appstore library 11-06)Doesn't work in Korea; Korean sites block foreign traffic heavily. Pickle uses Korean proxy IPs and local connectors.
OpenAI dotsBig-company features2026-09 baseNot verifiedWork-leaning always-on agent (research, coding, long-running projects); doesn't focus on Korean lifestyle services or phone calls
Claude DuetBig-company features2026-10-21 research previewNot verifiedAlways-on agent, not built for Korean local services
Grok BotiOS2026-08-11US rating 20k; Korea rating 271, Productivity #11User reviews say it failed to reschedule by phone while Pickle succeeded; no Korean local connections
ManusiOS2025-03Korea rating 3,225, Productivity #22General-purpose agent, no phone calls, no Korean services
모두의 AI (government + SKT / KT / Kakao)Web2026-10 betaTarget 15M users by year-endFree national project, but the beta leads with a chatbot; task-executing agents come later
Pally / Toki / InstinctOther2026Not verifiedOverseas product, runs through iMessage, Telegram or scheduling; no Korean localization
Where users come from
  • Verified founder Threads posts + closed beta by invite: 10-13 announced the Korean beta, 564 likes, 998 replies (mostly requests for invites); 10-22 and 10-26 posted updates, 200–300 likes each (TikHub scrape of @danifestoo)
  • Verified referral link loop: sign-up via pickle.com/i/… invite, capped invites per person; someone wrote a guide (foxcg 10-29)
  • Verified Threads seeding and word of mouth: a one-star review "Threads virality is the worst"; the founder posted a long thread about AI getting his subscription fee refunded
  • Verified word of mouth among friends: reviews like "I installed it for my mom, dad and friends too" and "spreading it to people around me"; the agent-to-agent scheduling feature pulls friends in
  • Verified press tests: Digital Daily 11-03, Digital Today 11-04, Game View 11-06, all after the chart surge
  • Inferred paid ads: nothing found in the metaads, tiktokads or googleads libraries (not on the tracking list)

Pickle: Your Own AI Agent (Pickle), and why it took off

App Store id6795790129 · developer Pickle Company (Pickle, Inc., Delaware, YC W25) · Google Play com.pickle.client · research date 2026-11-06. Store library history only goes back to 2026-10-28.

Bottom line

A Korean founding team in Silicon Valley has rebuilt its US personal AI agent (it calls and orders for you, books tickets, and schedules with friends' AI) as a Korean local version. Korean users were being flooded with US products like Meta's Muse, OpenAI's dots and Claude Duet, but couldn't use them: Muse isn't listed in the Korean store, and US agents can't get into Korean websites. Pickle caught this window and built momentum through the founder's Threads posts, a closed beta and a referral-link chain, reaching #3 overall in Korea's free iPhone chart and #1 in Productivity within three weeks. The growth is real: it's not inflated, and it's not an old product with a new launch date. But it's early: retention and payments are unproven.

1. Data: how big it really is

MetricValueSource
Korea iPhone free overall chart10-28 #23 → 10-30 #5 → 11-06 #3appstore library app_day, 2026-10-28–11-06
Korea Productivity Free chart10-28 #4 → 11-06 #1Same as above
iOS rating count (Korea)32 → 56 (9 days +24); Japan 1 → 11Same as above
JapanProductivity Free chart #35–64, never reached the overall chartSame as above
Google Play installs10-29 800 → 10-31 7,392 → 11-02 22,529 → 11-06 29,943 (7 days +29,143)googleplay library app_day / app_profile, 2026-11-06
Google Play rankKorea Productivity Free #22 → #6; Korea App Free overall #188 → #37; Korea Productivity Top Grossing #94 → #73googleplay library chart_rank
Android launch2026-10-19googleplay library app.released
third-party rankingMobileIndex-basis App Store daily chart: 10-26 #51 → 11-05 #3게임뷰 2026-11-06
company self-reported"Closed beta for 3 weeks, 100k active users, #1 in App Store Productivity"Founder Park Daniel (@danifestoo) Threads post, 2026-11-05,619 likes / 193 replies (scraped via TikHub). Self-reported, unverified

Why the rating count doesn't match the rank: it's #3 on the overall free chart with only 56 ratings. Our guess has two causes: (1) Korean users rarely leave ratings to begin with; (2) it's a closed beta, so many people downloaded it but couldn't get an invite and never got in, and so never rated. Google Play gained about 30k installs over the week, which matches the overall chart rank, so the downloads look real. The climb was gradual, with lots of media hands-on coverage over the same period that corroborates it. We see no sign of inflated numbers.

Is it an old product with a new date? No. The iOS 2026-08-10 is the date this global app first launched (the Japan store lists it as "Pickle (ピクル)", same id as the Korean store). The US version 10-01 was announced on X (Daniel Park X post), and the Korean version opened closed beta 10-13. The company has pivoted several times: it built an AI stand-in for Zoom, a desktop assistant, and AR glasses called "Pickle 1 / soul computer", which drew accusations of vaporware (SF Standard 2026-01-13). This agent app is its latest form.

2. Competitors: who else does the same thing

ProductWhereLaunchedScaleHow it differs from Pickle
Muse from MetaiOS/Android2026-09-08US store free overall #2, 460k ratings, 7 days +69k; Japan store #1 (appstore library 11-06)Not listed in the Korean store (no kr row in the library; Digital Daily: "gradually opening in the US"). The founder said: "Muse-style agents get blocked a lot when they hit Korean sites"
OpenAI dotsBig-company features2026-09 baseNot verifiedWork-focused: a cloud computer connects to 4000 apps for research, coding, and managing long-running projects; doesn't target Korean local services
Claude Duet(Anthropic)Big-company features2026-10-21 research previewNot verifiedAlways-on agent, likewise not aimed at Korean local services
Grok Bot(X)iOS2026-08-11US store rating 20k; Korean store rating 271, #11 in Productivity (appstore library)A user review says Grok Bot failed to make a reservation call using aitocall and tello, but Pickle got it done first try (10-18 reviews)
ManusiOS2025-03Korea rating 3,225, Productivity #22General-purpose agent; doesn't connect to Korean services or make phone calls
Modu AI (Korean government + SKT / KT / Kakao)Web / App2026-10 closed beta, official launch at year-endTarget of 15M users by year-end (NewsPim)National project, free. The beta starts with a chatbot; agents that can take actions come later (Global Economic)
Pally / Toki / InstinctOverseas apps / messaging channels2026Not verifiedPally works through iMessage and Telegram with near-free basic features; Toki focuses on scheduling; neither is localized for Korea
ChatGPT / Claude / Gemini AppiOSAlready listedKorean store Productivity chart #1–3Can chat, but can't call on your behalf or place orders on Korean websites

Bottom line: there are plenty of competitors and big companies are all building in this space, but as of 2026-11 early, no one in the Korean market could use Korean to finish tasks on Korean websites and over the phone for you. Muse, the strongest rival, isn't listed in Korea; Pickle grew into this gap.

3. Where users come from

ChannelEvidenceConfirmed?
Founder's Threads posts + closed beta where "comments / DMs need an invite"A 10-13 post announced the Korean iOS closed beta, 564 likes and 998 replies, most replies asking for invites; 10-22 posted an update (daily calls 10→20, quota ×1.5), 225 likes / 85 replies; 10-26 posted a security explainer, 306 likes / 80 replies (TikHub scrape of @danifestoo, 6,030 followers)Confirmed
Invite-link virality (scarcity)The closed beta requires signing up via a pickle.com/i/… invite link, with a cap on how many invites each person can send; someone wrote a guide on "how to get an invite" (foxcg 10-29); commenters asked for invites ("Jebra invite please" 10-14)Confirmed
Word of mouth and seeding on ThreadsA one-star review: "Threads viral is the worst" (spent a whole day on Threads before downloading, 10-29); the founder 11-01 posted a long post about getting a ₩40k subscription refund after the AI secretary 2 hours (82 likes / 66 replies)Confirmed
Word of mouth among acquaintancesComments: "I installed it for my mom, dad, and friends too", "I'm promoting it to people around me", "I'll spread the word around"; friends booking times with each other's Pickles is itself a way to pull people inConfirmed (comments)
Media hands-on testsDigital Daily 11-03, Digital Today 11-04, and Game View 11-06 each ran long hands-on reviewsConfirmed, but all came after the chart push; they amplified it, they didn't start it
Store charts10-30 steadily held #1–3 in the Productivity chart and reached the top 10 overall, bringing organic trafficInferred
Paid adsThe metaads, tiktokads, and googleads libraries show no Pickle advertisers or creatives (not on our tracking list; not finding it doesn't mean they didn't spend)Not found
TikTokSearching "피클 AI" only returns unrelated pickled cucumber videosNot a channel

4. Why it took off

  1. Timing: from 2026 9 to October, the US saw Muse (09-08), dots (late September), and Claude Duet (10-21) launch back to back; "AI that does things for you" became a hot topic, and Korea's homegrown "Modu AI" also began teasers in October. But Korean users had no agent that actually worked for them: Muse wasn't in the Korean store, and US products couldn't get past Korean website security plugins, PASS identity verification, or local payment methods. Pickle arrived in 10-13 and filled that gap.
  2. Actually completes Korea-specific tasks: connects to 60 domestic and foreign platforms, including government 24, the National Tax Service Hometax, Korail / SRT, Coupang, Naver, Kakao, and CatchTable (Digital Today). It uses Korean proxy IPs and local connectors, and hands back to the user steps that must be done in person, like PASS verification and payment. A reporter's hands-on test said "services other AI tools can't get into, Pickle can handle" (Game View).
  3. The phone-call "wow" moment: the AI actually places a call, orders at a restaurant, changes a flight, or calls customer service in Korean, with the call transcribed live. This is the thing most mentioned in comments ("it changed my plane ticket, which could only be changed by phone", "the owner looked really surprised"). It's easy to record and very shareable on Threads.
  4. Closed beta with invite links: the founder personally posted invites on Threads, turning scarcity into distribution. Nearly 1000 replies is itself free exposure.
  5. Free quota plus cheap reset tickets: there's a daily free quota that resets at 4 a.m.; once used up, you can buy reset tickets at ₩2,900 (half) or ₩5,800 (full) (Digital Daily). The barrier is low, and it's already monetizing: #73 in Google Play's Korean Productivity Top Grossing, in-app purchases $1.99–$200.

5. Differentiation, and whether that's why it broke out

  • Korea-specific execution (connectors, proxy IPs, handing authentication back to the user) plus real phone calls: this is the main driver. Comments and media tests both highlight "best optimized for Korea" and "actually finishes tasks on Korean websites", which is what they compare against Muse and dots.
  • Social among agents: your Pickle chatting with a friend's Pickle to schedule a time; the founder said "we're the only ones who've got it right worldwide." Media tests mention it, but comments mention it rarely. Our read is that for now it helps bring in new users more than it's the main reason people come.
  • Cute 3D character and friendly tone: some users like it ("it talks so prettily"), a bonus rather than a core feature.
  • Privacy pitch (TEE secure enclave, source code publicly verifiable): the founder brought this up in response to criticism to ease user concerns; it's not a growth driver.

6. What others can learn

  • The gap where "US big-company products get popular first but aren't usable locally": in language markets where popular overseas AI products haven't launched or can't work (Korea, Japan, Turkey, etc.), building a version that connects to local websites, local authentication, and local payments captures demand that big companies have already educated the market about.
  • Turning the demo into shareable content: AI actually calling and the shop owner being startled, told in a short video or long post, gets reshared naturally.
  • Founder-led closed beta with invite distribution: invites on Threads and X, where the founder has influence, make cold start free.
  • Charging per action: free daily, with cheap reset tickets when you go over, lowers the barrier compared with a subscription.

7. Still unclear

  • "100k active" is self-reported by the founder, and how "active" is defined is unclear. Google Play shows only 30k installs; iOS download counts aren't available.
  • Retention: reviews complain a lot: it's slow (booking one KTX ticket takes 20 minutes), Coupang login fails, the quota is too small, outbound calls are flagged as spam, and logging in from an overseas IP triggers Naver security alerts. How many of the people who pushed it up the charts will stick around is unknown.
  • Revenue: iOS hasn't reached the Top Grossing chart, and Google Play Top Grossing only got to #73, so paid scale is probably still small (our inference).
  • How much moat is left once Muse launches in Korea, or once "Modu AI" adds agent features in its December full release.
  • Japan version just getting started (Productivity chart #35–64). Whether it can copy Korea's playbook: no signal yet.
  • Funding: the startup database says roughly $26M raised in total (Craft, FirstMark, First Round, YC). Not verified against a primary source.

Sources

This article was written by AI from public data and the data we collect daily. Numbers are sourced, and inferences are labeled.