# MYMI, a Korean AI character chat built by two people: 5,500 installs got it onto a Top Grossing chart, but the chart bar is lower than you'd think · Mobile app

> Source: https://stringers.io/en/apps/play_live.mymi.app.html (Stringers / 什么值得做, generated 2026-11-08)
> Stringers answers one question every day: what small product can one or two people build in a few months and actually make money from. Numbers come from 15 databases we collect daily, each with source and date.
> For agents: these verdicts and the underlying databases (verified revenue, App Store / Google Play charts, ad libraries, GitHub, Product Hunt…) are queryable live over MCP — tools today / hot / products / directions / direction / data. Setup: https://github.com/komako-workshop/stringers (the user logs in at stringers.io → "配置 agent" to get a personal URL). Site index: https://stringers.io/llms.txt

> Not really big. Through transparent price lists, uncensored adult content and fast model updates, it kept a small group of heavy users who chat for hours a day and spend 100k won over two weeks. It recruits through posts in the DCInside community, points giveaways for word of mouth, and app-exclusive lotteries. No paid acquisition.

[Mobile App](https://stringers.io/en/apps/index.md) · Android · written 11-6

MYMI - AI Character Chat & RP · WonMo Inc.

# MYMI, a Korean AI character chat built by two people: 5,500 installs got it onto a Top Grossing chart, but the chart bar is lower than you'd think

**What it is** An adult-oriented AI roleplay chat website and app from a two-person Korean company. Each reply deducts points based on the chosen model, and topping up on the web is cheaper than in the app.

**Why it took off** It hasn't really taken off. Transparent pricing, uncensored adult content, and quick adoption of new models keep a small group of heavy users who chat for hours a day and can spend ₩100k in two weeks. Acquisition runs on posts in the DCInside community, giveaways of points for word of mouth, and an app-exclusive lottery, not paid ads.

**C** **How you could approach it** Don't launch the fifty-somethingth Korean clone. Borrow the small-team playbook instead: take payments on the web and use the app as an entry point, publish pricing and refund rules, trade points for word of mouth and invites, post official threads in communities, and build SEO on work pages. Then apply it in a language market where people already run RP on their own APIs but no local platform exists yet.

Grade C: Korean AI character chat already has two leaders, Zeta and Crack, plus fifty-plus followers. MYMI is a trailing follower itself. Its features are common and users churn easily. Its acquisition playbook is worth studying, but copying it wholesale isn't worth the effort. Few people know about it yet

Data (updated daily) · first launch 2026-07-22 · 11-07 installs 5,572

Android · Total installs worldwide

Data 2026-10-29 → 2026-11-07 (10 days, 10 observations). Google's detail page only refreshes install counts every other day, so a flat day doesn't mean nobody installed.

WonMo, a Seoul company with just 2 employees, built MYMI, a Korean AI character roleplay platform that charges per message and lets users choose Claude, Gemini, Grok, or GPT for their chats. On Google Play it has only 5,572 installs yet sits at #131 on the Korean Entertainment Top Grossing chart. At first glance it looks like a quiet business with few people and lots of money. Look closer and it's a late entrant in a niche already crowded with fifty-plus competitors. Below: what it is, how it keeps paying users hooked, where its users come from, and what this means for you.

### A newcomer who spent ₩100k in two weeks

On the evening of Oct 10, 2026, at 11 p.m., someone on the DCInside AI chat board asked for recommendations for "a beginner-friendly platform". In the post he explained his experience: he first saw an app in an Instagram ad, couldn't figure it out, cursed at it, then switched to MYMI. "I spent ₩100,000 and had a great time for about two weeks", chatting even while slacking off at work, "I felt like I was holding onto it for 8 hours a day". His last line: "Sparks drain too fast… it's scary. I'm going to wrap up the storyline I have going and switch to another app." (Post views: 758)

Those few lines capture MYMI's business: it can cost one person about $70 in two weeks, the money drains per message and fast, and users can leave at any time.

### What MYMI is

The developer is WonMo Inc. (원모 주식회사), registered in 2023 December, with 2 employees. It previously made two small diet and calorie-tracking apps, each with only one or two reviews (Korean company registry and App Store). MYMI first launched on the web in 2025 June as "AI characters you can video-chat with", then changed to the text roleplay it has now. I couldn't find out why it pivoted.

Current form:

- **The website is the main front** : mymi.live supports Korean, English, Japanese, and Traditional Chinese, and the Korean site has roughly 1,900 creator-made characters. The Android version is only listed in the Korean store (store date shows 2026-10-25). The iOS version went live in Korea and Taiwan in September and has only 6 reviews.

- **Per-message Spark deduction, with the model chosen by the user.** The website's pricing table (11-04 pricing) is detailed: the in-house MYMI Lite costs 20 Spark per reply, the default Gemini 3.1 Pro costs 120, and Claude Opus 5.5 costs 300. Replies over a length threshold carry a surcharge per 100 tokens. Points are pre-charged and refunded based on actual length. Model refusals deduct only 20%, and errors are fully refunded. On the web, 10k Spark sells for ₩10k, so one Opus reply costs about ₩300 (roughly $0.2).

- **Adult-oriented.** The homepage slogan is "검열 없이 자유롭게" (freely, without censorship). Users certified as 19 or older can turn off the safety filter. Genres include fantasy, romance, BL, and wuxia.

- Long-term memory, visual novel mode, and AI-generated scene images are all standard across the category.

- You can chat 5 times before signing up.

By the pricing above, the ₩100k spent by the newcomer at the top of this piece equals roughly 330 Opus replies, or 800 or more replies on the default model (inferred from web pricing). Chatting 8 hours a day and running through that in two weeks is entirely plausible.

### Installs versus earnings

Cumulative Google Play installs: 10-29 is 4,657,11-06 is 5,572,7 days +915. Growth comes in steps: 10-31 +322, 11-03 +505, 11-05 +88, and the other days are 0. Since launch, it averages about 36 installs per day (third-party estimate).

It has charted on only one list: the Korean Play Entertainment Top Grossing chart, where it sat at #10-29–11-06 from 128 to 135. On iOS it ranked #109–125 in Korea's Entertainment Top Grossing and #128–194 in Taiwan.

**Don't take this chart at face value.** On the same day (11-06), 벨벳톡 ranked #128 with only 752 installs, and 티팟 at #129 had 1,103. The bar for the top hundred or so on Korea's Entertainment Top Grossing chart is low. So "5,500 installs gets you onto the chart" proves that individual payers do spend real money, but absolute revenue is probably small, on the order of a few hundred thousand won per day (estimate, unverified).

On the other hand, the Play chart may understate it: topping up on the web is about 17% cheaper (10k Spark costs ₩10k on the web and ₩12k on Play and the App Store, per the official payment guide), and experienced long-term users pay on the web. Web revenue is invisible from outside, and it may well be the bulk of the business.

### Where users come from: almost no ad spend

**Two DCInside boards.** MYMI posts officially on the AI chat board and the AI chat women's board. 2025-06-06 The first post, "Vivid video chats! New AI chat service MYMI", got 2,668 views. It was followed by a Gemini pass campaign (2026-05), the launch of scene snapshots (10-22), a Grok dice event (10-23), and "Promote MYMI and get Grok" (10-25). Each post got tens to hundreds of views. On the women's board, a user compiled a summary of check-in reward expiry dates across platforms, and MYMI is listed (693 views).

**App-exclusive events.** The "Grok lucky dice" that started on 10-23 can only be rolled in the app and requires identity verification. It awards up to 30k Grok credits. The two Play install steps (10-31 +322, 11-03 +505) fall right after the event. That the event drove these installs is an inference.

**Trading points for word of mouth.** "MYMI 소문내기" (spread the word about MYMI): publicly post on social platforms or blogs with #MYMI, and receive up to 8,000 Grok. Invite a friend, and once they make their first top-up, you each get 1,500 Spark. What it pays out isn't cash; it's a bit of inference cost.

**SEO and free trials without signup.** Every work, tag, and creator has a landing page that reads "가입 없이 바로 시작" (start right away, no signup). The official blog ran a dense series from 10 to November on questions like "How to choose an AI character chat app" (11-03). Search the brand name and the top results are its own sites.

**No paid acquisition in sight.** The site has an X ad conversion pixel, but nobody has mentioned MYMI on X in 90 days. Not in the Meta ad library, nothing on TikTok, Instagram at 67 followers, YouTube at 2 subscribers, Discord at 576 members, 25 online. No results doesn't mean no ads, but there's no sign of large-scale spend.

### Who it's competing with: Korea already has over fifty

AI character chat in Korea is already an established category. 11-06 In the same Korean Google Play Entertainment Top Grossing top 200, there are more than 50 similar apps.

| Competitors | Scale (Google Play) | Compared with MYMI |
|---|---|---|
| Zeta(Scatter Lab) | 7.72M installs, 7 days +297k, #6 in Entertainment Top Grossing; 2025 revenue about ₩26.7B (Chosun Ilbo 10-13) | Category leader, in-house models, skews younger |
| Crack(wrtn) | 1.14M, Entertainment Top Grossing #4 | Also sells high-cost models like Claude Opus by the point. Two orders of magnitude bigger. |
| 위프(Bunkerkids) | 1.6M, launched only when it reached Top Grossing #14,2025-10 | Aimed at women, BL web-novel style chat. Positioning closest to MYMI's recent direction. |
| BabeChat(베이비챗) | 170k, #32 bestseller | Also Gemini / Claude multi-model plus adult mode, nearly the same model |
| Melting, Frontia, Bagelchat, Tikita, Lipin, Mood, Kkuuk, Dumchung, Tpot, etc. | A few thousand to 400k | Undifferentiated, each with a small hook like progression or romance sims |
| Character.AI, PolyBuzz, Emochi | 70.7M / 57.9M / 29.7M | Global majors; in the Korean bestseller charts they rank 169, 63, 103 |
| Players bringing their own API key on arca.live | Unknown | Power users wire up Claude / Gemini themselves; MYMI packages that setup as a product and sells it per message |

MYMI didn't invent this and doesn't lead on any metric. Its transparent price list and refund rules are more detailed than Crack's or BabeChat's, which helps keep price-comparing heavy users, but nobody else couldn't do it.

Regulation is tightening too: October big players were called before Congress over teen issues and began adding limits of 3 hours a day and ₩100k a month (ZDNet 10-12). That only covers minors and has limited impact on adult-oriented platforms like MYMI, but it shows the category is now under watch.

### What's keeping it alive

- **The track is still growing, and the long tail gets a slice.** Zeta's revenue 2025 was more than 5× the year before, and users average 49 hours a month in the app (Chosun Ilbo / 모바일인덱스). Heavy users the top players can't satisfy spill over, looking for places that are freer and offer more models.

- **Heavy users pay a lot.** The top web tier is ₩100k, and the highest in-app purchase is $89.99.

- **New models ship fast.** Claude Opus 5.5 and Gemini 4 Argon are already on the price list, and heavy RP users care about that.

- **Near-zero acquisition cost.** Community posts, word of mouth via points, in-app events, plus SEO and free trials without signup. Two people can run it.

These explain why people pay. They don't explain scale, because scale never materialized.

### If it were you

**C: worth studying, not worth copying in Korea.** Korean AI character chat already has two leaders and over fifty followers. MYMI is itself a follower, and it's behind. Multi-model, points-based, adult, long-term memory: everyone has these. Entering now means being the fifty-something-th, with no real spot. These users also switch platforms easily; the person at the top of this thread is paying while planning to leave.

**What's worth taking is this small team's playbook** : collect payment on the web, use the app as an entry point; publish the price list and refund rules; use points to earn reviews and invites; post official updates consistently in vertical communities; build landing pages for every work, tag, and creator for search.

**Who it fits:** small teams that already have a character chat or AI companion product and want to lower acquisition costs; people who understand a specific minority-language RP community (with its own forums, where people already play using their own API keys).

**Who it doesn't fit:** anyone trying to enter Korea from scratch; anyone who can't handle adult-content moderation, identity verification, and payment risk controls; anyone counting on a chart ranking to raise money (the bar for a hundred-plus rank is low).

**How to do it:** if you're following, change the market, not the features. Find a place where people in a local language already connect APIs to play roleplay, but no local platform exists yet (which languages fit, I didn't check; you'd need to look at local forums yourself). Build around "multiple models to choose from, per-message pricing, few refusals charged, refunds on errors." Web first. Web prices lower than the app, which saves you the store commission.

**How to acquire users:** copy MYMI: post officially on local AI chat forums; use inference credits rather than cash rewards to reward public "share a post" and "invite a friend who tops up"; run app-only lotteries to move web users into the store, driving installs and chart position; let people chat 5 times without signing up.

**Deadline:** there's no hard deadline, but each market's window is shrinking. Once big players like Character.AI or PolyBuzz localize, the gap in minor-language markets closes.

**What would show you're wrong:** (1) MYMI's Play installs keep rising by thousands per week for several weeks and it climbs into the Top 50 Grossing, which would show that "transparent pricing + adult" can break through a crowded category by itself, and that its approach is worth copying outright; (2) it has published its web revenue, and it's clearly an order of magnitude above the estimated "hundreds of thousands of won a day," which would show the web-payment route is badly underestimated; (3) conversely, if Play installs flatten once the dice event ends, the steps were just manufactured by the event.

### Data and sources

- Installs and charts: Google Play Korea, App Store Korea and Taiwan, 10-29 to 11-06; daily installs are third-party estimates.

- Product and pricing: mymi.live homepage, model price guide (11-04 pricing, 11-06 read), payment guide, FAQ, blog. Not verified with an actual top-up.

- Company: Korean business registry data (bizno.net).

- Community: posts from DCInside AI chat board 2025-06-06, 2026-05-01, 10-10, 10-23, and the women-oriented AI chat board 10-06.

- Competitors and industry: Chosun Ilbo 2026-10-13, ZDNet 2026-10-12, Namu Wiki Crack entry, Google Cloud BabeChat case page; competitor install counts are from Google Play 11-06.

- Anything marked "inferred," "estimated," or "not found" is an inference or wasn't verified.

[Android store page ↗](https://play.google.com/store/apps/details?id=live.mymi.app)

**Research notes: competitor table, acquisition evidence, full research report**

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| Zeta(제타,Scatter Lab) | Android | 2022-05 | Play 7.72M installs, Korea Entertainment Top Grossing #6; annual revenue about ₩26.7B in 2025 | Category leader, in-house models, skews 10–20 years old; MYMI offers multiple third-party models, web first, aimed at adults |
| Crack(크랙,wrtn) | Android | 2025-03 | Play 1.14M, Korea Entertainment Top Grossing #4 | Also sells high-cost models like Claude Opus by the point, with two orders of magnitude more scale; MYMI's price list and refund rules are more detailed |
| 위프(Bunkerkids) | Android | 2025-10 | Play 1.6M, Korea Entertainment Top Grossing #14 | Also aimed at women / BL web-novel style chat, with the closest recent positioning, and far larger |
| BabeChat(베이비챗) | Android | 2025-02 | Play 170k, Korea Entertainment Top Grossing #32 | Also Gemini / Claude multi-model Pro mode plus adult mode. MYMI is nearly the same model. |
| Melting / Frontia / Bagelchat / Tikita / Lipin / Mood / Kkuuk / Dumchung / Tpot, etc. | Android | 2025–2026 | Play from thousands to 400k, all in Korea Entertainment Top Grossing top 150 | Undifferentiated Korean AI character chat, each with a small hook like progression, romance sims, or short dramas |
| Character.AI / PolyBuzz / Emochi | Android | 2023–2024 | Play 70.7M / 57.9M / 29.7M | Global majors; weaker on Korean content and adult freedom than local platforms |
| The arca.live BYO-API-key approach | Other | — | Community size unknown | Power users connect Claude / Gemini APIs themselves. MYMI productizes that setup and sells it per message. |

Where users come from

- **Verified** **official posts on the DCInside AI chat board and women-oriented board** : 2025-06-06 first-post views 2,668; 2026-05 Gemini pass, 10-22 snapshot, 10-23 dice, 10-25 promo credit giveaway and other official posts, with views in the dozens to hundreds

- **Verified** **Community word of mouth** : a DC 2026-10-10 post has a user saying they switched to MYMI from another app they saw in an Instagram ad, spending ₩100k in two weeks. A roundup thread of check-in rewards for female-oriented apps lists MYMI.

- Inferred **app-only event funnel** : from 10-23, the Grok lucky dice could only be rolled in the app and required identity verification, with up to ₩30k in credits; afterward Play 10-31 +322, 11-03 +505

- **Verified** **points for UGC / invites** : posting publicly with #MYMI earns up to 8,000 Grok; inviting a friend who tops up gives both sides 1,500 Spark

- **Verified** **SEO and free trial without signup** : works, tags, and creators all have landing pages; the blog publishes problem-focused articles densely from 10 to November; users can chat 5 times without registering

- Inferred **X ads** : the site has an X ad conversion pixel; the X corpus shows 90 days of 0 discussion

- Inferred **Meta / TikTok paid acquisition** : the metaads library has no record of this advertiser, TikTok has no relevant content, Instagram 67 followers

## MYMI - AI Character Chat & RP (live.mymi.app) Research Report

> Channels: Google Play (Korea store only) + iOS (Korea, Taiwan) + web at mymi.live (the main battlefield). Developer: WonMo Inc. (원모 주식회사). Research date 2026-11-06.

### Bottom line

**It isn't really "breaking out"; the recall score overrates it.** MYMI is a Korean-language AI character roleplay chat platform built by a 2-person company in Seoul. 2025 June it launched on the web as "video AI character chat", then pivoted to "multi-model, per-message points" text RP. It's squeezed into a Korean category that's already crowded (in the same Korea Play entertainment Top Grossing top 200, more than 50 similar AI character chat apps show up), ranked #131, with small apps at a few hundred to a thousand installs sitting right next to it. Two things are worth a look: first, it earns a high per-person payment from a small group of heavy adult users, with payment on the web and the app serving only as an entry point; second, its user acquisition relies almost entirely on DCInside's AI chat board, points-rewarded UGC tasks and an "app-exclusive dice roll" event, with no paid acquisition.

### Data (from our library)

| Metric | value | Source |
|---|---|---|
| Cumulative Play global installs | 4,657(10-29) → 5,572(11-06), +915 in 7 days | googleplay.app_snap |
| Play daily gains | 10-31 +322,11-03 +505,11-05 +88, other days 0 (observations are stepwise) | googleplay.app_day |
| Average per day since launch | About 36 installs/day, starting from only 2026-06-07 installs 10 (AppstoreSpy estimate, averaged) | googleplay.app_day_ext |
| First review / store launch | 2026-07-22 / 2026-10-25 (store dates were reset) | googleplay.app |
| Play charts | Only on the Korea entertainment Top Grossing chart, ranked 128–135 from 10-29 to 11-06 | googleplay.chart_rank |
| iOS | Listed 2026-09-13, 6 ratings; Korea entertainment Top Grossing 109–125, Taiwan entertainment Top Grossing 128–194 | appstore.app_day |
| In-app purchases | $2.29–$89.99 per item | Play listing |
| Company | WonMo Inc., registered 2023-12-06, 2 employees, CEO 김정원 | bizno.net |
| Same developer | Dietary (weight-loss coach, iOS 2024-10), Joule AI (food calorie tracking, 2024-11), each with only 1–2 reviews | appstore.app_profile |

**How to read these numbers** : ranking #131 on the Korea Play entertainment Top Grossing chart is a low bar. On the same day, the #128 app 벨벳톡 had only 752 installs, and the #129 app 티팟 had only 1,103 installs (googleplay.v_chart 2026-11-06). So the signal "5,500 installs gets you onto the Top Grossing chart with unusually high ARPU" only half holds: per-person spending is indeed not low, but absolute revenue is most likely small. Exact figures are unverified; the order-of-magnitude estimate is tens of thousands of won per day. Also, web top-ups are about 17% cheaper than in the app (10k Spark on the web for 10,000 won, 12,000 won on Play / App Store, per the mymi.live payment guide), and savvy long-time users pay on the web, so the Play chart understates its revenue.

### Competitors (doing the same thing)

AI character chat / interactive fiction in Korea is already a mature category, with two leaders, dozens of followers, and regulatory pressure:

- **Zeta (제타, Scatter Lab)** : 7.72M installs on Play, +7k over 297 days, #6 on the Korea Play entertainment Top Grossing chart; annual revenue of about 2025B won in 26.7 (Chosun Ilbo 2026-10-13).

- **Crack (크랙, wrtn)** : 1.14M on Play, #4 on the Korea entertainment Top Grossing chart; also offers premium models such as Claude Opus, billed by points (나무위키, crack.wrtn.ai announcements).

- **위프 (Bunkerkids)** : 1.6M on Play, launched 2025-10, #14 on Top Grossing, focused on women-oriented / BL web-novel-style chat, closest to MYMI's positioning and far larger.

- **멜팅, 프론티아, 베이글챗, 케이브덕, 베이비챗 (BabeChat), 티키타, 리핀, 무드, 꾸욱, 둠칫, 티팟, 캐릭터톡 (Postype)** and others: installs range from a few thousand to hundreds of thousands, all on the same Top Grossing chart (googleplay 2026-11-06). BabeChat also offers Gemini / Claude multi-model options (Google Cloud case study page, channel.io docs).

- Overseas majors: Character.AI (70.7M on Play), PolyBuzz (57.9M), Emochi (29.7M), ranking #169, #63, #103 respectively on the Korea Top Grossing chart.

- Community DIY: the AI chat board on arca.live has many people running Claude / Gemini with their own API keys, another route for heavy users.

**Bottom line** : MYMI is not the category pioneer and shows no clearly leading metrics; it's one of the lower-ranked players in this category.

### Differentiation: yes, but not unique

1. **A transparent "per-message × per-model" price list** : one reply costs 20–300 Spark; the in-house MYMI Lite / Flash / Plus are the cheapest, Claude Opus 5.5 / 4.6 the most expensive, plus Gemini 4 Argon, Grok 4.3, GPT-5.4. Billing follows actual output tokens with true-up, model refusals charge only 20%, and errors are fully refunded (mymi.live/guide/chatting/ai-models, 2026-11-04). Crack and BabeChat are also multi-model plus points; MYMI's difference is a finer price list and clearly stated refund rules, which appeal to budget-conscious heavy users.

2. **Adult / "uncensored" positioning** : homepage tagline "검열 없이 자유롭게" (free, without censorship). After real-name verification, users aged 19+ can turn off the safety filter for adult works (FAQ). Major players 2026-10 were called out by parliament over teen issues and added a daily cap of 3 hours and a monthly limit of 100k won (zdnet 2026-10-12), but that only covers minors and has limited impact on adult users.

3. **Long-term memory (memory notebook / memory fragments), visual novel mode, AI scene snapshots** : comparable products mostly have these too, so they don't form a moat.

4. **Web first, one account across platforms** , supports Korean, English, Japanese and Traditional Chinese, and payment on the web is cheaper.

**Are these differentiators why it "broke out"?** They only explain why it keeps a small set of users willing to pay; they don't explain scale growth, because scale never really took off.

### Growth tactics (with evidence)

| Channel | Evidence | Confirmed? |
|---|---|---|
| Official posts on the DCInside AI chat board and the AI chat women's board | 2025-06-06 first post "영상으로 생생하게! 신규 AI 채팅 서비스 MYMI" (Video makes it vivid! New AI chat service MYMI), 2,668 views; 2026-05-01 Gemini pass giveaway post, 412 views; 10-22 snapshot feature, 10-23 Grok dice, 10-25 "promote MYMI and get Grok" posts, each getting only dozens to a few hundred views | Confirmed |
| User word of mouth | Newbie recommendation thread on the AI chat board 2026-10-10 (758 views): a user said they switched to MYMI from another app advertised on Instagram, "spent 100k won in two weeks, playing about 8 hours a day"; a roundup post on the women's board about check-in rewards across platforms included MYMI (693 views) | Confirmed |
| App-exclusive event drives traffic | Starting 10-23, the "Grok lucky dice" can only be rolled in the app, and requires real-name verification first, with up to 30k Grok credits. Play installs +10-31 and +322, +11-03 and +505 fall right after the event | Event confirmed; driving traffic is an inference |
| Points for UGC | "MYMI 소문내기" (spread the word about MYMI): publicly post on social media or blogs with #MYMI for up to 8,000 Grok; after an invited friend's first top-up, both sides get 1,500 Spark each | Confirmed |
| SEO | Every tag, creator and work has a landing page saying "가입 없이 바로 시작" (start right away, no sign-up); the blog publishes tutorials, picks and recommendation posts densely from 10 to November, such as the 11-03 post "how to choose an AI character chat app"; the top results for brand-name searches on Jina are all the official site | Confirmed |
| Free trial without sign-up | Users can chat 5 times without registering (Gemini 3.8 Flash), lowering the barrier to a first try | Confirmed |
| X ads | The official site has an X (Twitter) ad conversion pixel; in our X corpus there was no MYMI discussion for 90 days | Pixel confirmed; ad buying has no evidence |
| Meta / TikTok | No such advertiser in the metaads library; no related content found on TikTok; Instagram has 67 followers, YouTube 2 subscribers | No ad buying found (not finding it doesn't mean it didn't happen) |
| Discord | 576 people, 25 online | Confirmed |

### Why it has "some momentum"

- The category itself is growing: Zeta's revenue for year 2025 was more than 5 times the prior year, and users average 49 hours a month (Chosun Ilbo / MobileIndex), so long-tail platforms can catch the heavy users spilling over.

- Heavy users pay a lot per person: the web has a tier at 100k won, the app tops out at $89.99, and one user self-reported spending 100k won in two weeks (DC 2026-10-10).

- Acquisition costs almost nothing: community posts, points for content, app-exclusive events, combined with SEO and a no-sign-up trial, which fits the cost structure of a 2-person team.

- It adopts the latest models quickly: Claude Opus 5.5 and Gemini 4 Argon arrived fast, which matters to heavy RP users.

### Visibility

**Still few people know about it.** No media coverage, 0 discussions on X, and no public revenue figures; it's only mentioned in the small circle of the DCInside AI chat board. But the whole category has been widely covered in Korea (parliamentary inquiries, revenue news), so the "Korean-language AI RP" path is already crowded.

### Copycats

No one clearly copying MYMI has turned up. It is itself a follower of the Zeta / Crack / BabeChat model.

### What to borrow

- "Charge on the web, use the app as the entry point": the web is about 17% cheaper, one account shared across platforms; the app uses an exclusive event (dice roll) to drive installs, treating store charts and install counts as a visibility channel.

- Publish the model price list and refund rules clearly (refusals charge only 20%, errors fully refunded); for heavy RP users, that is itself a selling point.

- Use points instead of cash to reward UGC and invites: public #MYMI posts earn credits, and both sides of an invite get a reward after the first top-up. The cost is marginal inference, not ad spend.

- Keep posting feature and event updates in vertical communities (DC's AI chat board and women's board) under an official identity.

- Programmatic SEO: one page per work, tag and creator, plus question-style blog posts like "how to choose a chat app" and "why AI forgets things".

### Still unclear

- Real revenue: the amounts behind the Play / iOS top-grossing ranks are unverified, and web payments are completely invisible and may be the bulk of it.

- Web traffic and sign-ups (Korean-language site 1,900 characters, but the number of active creators and users is unclear).

- Whether the X pixel really runs X ads, and how much is being spent.

- Whether the install lift came from the campaign or from random fluctuation (only 9 days of observations).

- Why the video-chat version (2025-06) didn't work out, and when the direction changed.

### Sources

- googleplay / appstore library (agent-im server, 2026-10-29 to 11-06)

- [www.mymi.live/ko](https://www.mymi.live/ko) ,/guide/chatting/ai-models ,/guide/payment/payment-methods ,/guide/faq ,/blog ,/blog/choose-ai-character-chat-app

- [bizno.net/article/1518603288](https://bizno.net/article/1518603288)

- [gall.dcinside.com/mgallery/board/view/?id=aichatting&no=2469…](https://gall.dcinside.com/mgallery/board/view/?id=aichatting&no=24693) (2025-06-06)

- [gall.dcinside.com/mgallery/board/view/?id=aichatting&no=3278…](https://gall.dcinside.com/mgallery/board/view/?id=aichatting&no=327892) (2026-05-01)

- [m.dcinside.com/board/aichatting/394889](https://m.dcinside.com/board/aichatting/394889) (2026-10-23)

- [gall.dcinside.com/mgallery/board/view/?id=aichatting&no=3900…](https://gall.dcinside.com/mgallery/board/view/?id=aichatting&no=390025) (2026-10-10)

- [gall.dcinside.com/mgallery/board/view/?id=aiotomechatting&no…](https://gall.dcinside.com/mgallery/board/view/?id=aiotomechatting&no=20674) (2026-10-06)

- [zdnet.co.kr/view/?no=20261012104704](https://zdnet.co.kr/view/?no=20261012104704) ;[www.chosun.com/economy/tech_it/2026/10/13/KTPHEGDB7JGKRJRELB…](https://www.chosun.com/economy/tech_it/2026/10/13/KTPHEGDB7JGKRJRELBRZHS7NZE/)

- [namu.wiki/w/크랙](https://namu.wiki/w/크랙)(애플리케이션) ;[cloud.google.com/customers/intl/ko-kr/babe-chat](https://cloud.google.com/customers/intl/ko-kr/babe-chat)

- Discord invite endpoint (576 members); x-radar search for "mymi" over 90 days turned up 0 posts; TikHub search found nothing relevant; metaads library has no records

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