# Wedding photos with AI characters: a game-company-style launch pushed an AI chat app into the top 25 of Korean and Japanese entertainment charts

> Source: https://stringers.io/en/apps/play_com.dearbe.dearbe.html (Stringers / 什么值得做, generated 2026-11-20)
> Stringers tracks which small products suddenly took off each day, investigates why one by one, and distils playbooks from them. Numbers come from 15 databases we collect daily, each with source and date.
> For agents: these investigations and the underlying databases (verified revenue, App Store / Google Play charts, ad libraries, GitHub, Product Hunt…) are queryable live over MCP — tools feed (daily breakouts) / plays (playbooks) / hot / products / data. Setup: https://github.com/komako-workshop/stringers (the user signs in with Google at stringers.io/en/start.html to get a personal URL). Site index: https://stringers.io/llms.txt

> It likely has resources from listed game company Joycity behind it, and at launch it used the full game-publishing playbook: same-day press releases, a KRW 15M contest, sign-up rewards in tokens, poaching creators from rivals, paid creators and feed ads. The Korea and Japan Entertainment free charts have a low bar, so 20K installs were enough to rank top 20.

[Mobile App](https://stringers.io/en/apps/index.md) · Android · written 11-9

Dearbe: More Than AI Chat(디어비) · DEARBE Corp.

# Wedding photos with AI characters: a game-company-style launch pushed an AI chat app into the top 25 of Korean and Japanese entertainment charts

**What it is** : an app for chatting with AI characters. Beyond chat, it turns conversations into songs and illustrated novels, and uses your selfie with a character to composite couple photo strips and wedding photos, aimed at web-novel and fan-culture readers in Korea, Japan, and Taiwan.

**Why it took off** : there's likely a listed game company, Joycity, behind it, and the launch used the full game-publishing playbook: same-day press releases, a ₩15M contest, free tokens on signup, recruiting rivals' creators, paid influencers, and feed ad buys. Entertainment free charts in Korea and Japan have a low bar: 20k installs was enough to reach the top 20.

**C** **How you could follow it** : don't go head-to-head with AI companion platforms. If you do play, pick out its most profitable "merch" piece and build a web tool that generates character photos and custom songs for existing zeta and Crack users, charging per image or per song, and acquire customers through the finished outputs and referral codes within their communities.

Grade C: you can learn a few tactics from it: ad hooks, small contests with referral codes, and recruiting rivals' creators. But it's a latecomer backed by a game company, using publishing budget to push into the zeta/Crack/Talkie red ocean, and it hasn't proven paid revenue yet. A small team copying it won't get the same resources. Few people know about it yet

Est. monthly revenue (Google Play) **under $1K** We estimated this from its rank on country top-grossing charts and real install counts (it isn't on the grossing chart, so confidence is lower), calibrated against real revenue; typical error is 2–3x. Read it as a tier, not an exact number.

Data (updated daily) · first launch 2026-10-31 · 11-20 installs 39,993

Android · Total installs worldwide

Data 2026-11-04 → 2026-11-20 (17 days, 17 observations). Google's detail page only refreshes install counts every other day, so a flat day doesn't mean nobody installed.

Android · Ratings count (cumulative)

Data 2026-11-05 → 2026-11-20 (16 days, 16 observations)

Dearbe (Korean name 디어비) is an AI character chat app from a new Seoul company, and its pitch is "more than chat": it turns your conversations with characters into songs and illustrated novels, and uses one selfie to composite couple photo strips and wedding photos. It launched on Google Play and the web in November 2, reached about 9k installs in 21 days, and made the top 25 of the free entertainment charts in Korea, Japan, and Taiwan. The rankings look good, but nearly all of those 20k people were paid for: press releases, a contest, token giveaways, influencer ads, and paid acquisition, all of it. It's worth taking apart, but not worth copying.

### A "#ad" post

In early November, a Korean Instagram creator who covers AI, @prompt_what (프롬왓), posted something that opened with the hashtag "#광고" (ad):

> "You can even get wedding photos taken with AI characters 💍 디어비, the AI character chat app that launched in November 2, lets you make everything from couple photo strips to wedding photos using a character from your chat plus one photo of me."

The rest is standard affiliate-style copy: a global creator contest with a total prize pool of 15M won (about $10k USD), closing Nov 22; sign up before Nov 30 and get 1,000 Spark tokens; it installs from Google Play and also plays directly in the browser; last line: "Comment 『디어비』 below and I'll DM you the install and sign-up links."

A few days later, two one-star reviews appeared on the Google Play Japan listing. One, from November 4: "Saw this app in an ad, installed it, and the text was tiny, so I uninstalled right away." Another, from November 7: "Totally different from the ad. Not very fun. zeta has more characters worth talking to." Each had a dozen-plus "helpful" votes.

Put these two things side by side, and that's roughly what Dearbe is right now: the ads work, and the product hasn't caught up.

### What it is

- **Company** : DEARBE Corp., based in Mapo-gu, Seoul, with representative 배준석 (Bae Jun-seok). Its privacy policy says the listed game company Joycity (developer of Freestyle and Gunship Battle) runs its cloud infrastructure, membership login platform, and payment integration. When a startup hands accounts and payments to a listed game company, it's likely a subsidiary, incubator project, or deep partner (inference; equity not checked).

- **Chat** : the same conversation can be viewed as a chat log or in a "novel" format; over 35 genres: romance, BL / GL, time travel, regression, wuxia, idol, chaebol, clearly built to suit Korean web-novel readers.

- **Beyond chat** : conversations turned into songs (OST, with vocal and instrumental versions); turned into illustrated novels; selfie plus character for couple photo strips, wedding photos, magazine covers; fan-cam and challenge-dance videos of characters; collectible virtual figures of characters; characters post "Moments" and can take voice calls; tap different spots on a character and the expression changes in real time.

- **Creators** : a one-line concept generates a character draft, and becoming an official creator earns revenue share.

- **Monetization** : Spark tokens; check-ins and tasks hand out a few, the rest comes from in-app purchases, listed on Google Play at $0.79 to $77.90 per purchase; ads on top.

### How much it grew

| Date | Google Play cumulative installs | Score |
|---|---|---|
| 11-05 | 3,812 | 16 |
| 11-08 | 15,182 | 37 |
| 11-10 | 20,812 | 46 |

(Google Play store data. From 11-05 to 11-07 the count sat at 3,812, then jumped to 15,182 on 11-08. That's the store's install counter updating late, not a real day of 10,000+ installs. Counting from the 11-02 launch, that works out to about 2,300 installs a day on average.)

The chart positions were all in the Entertainment free charts: Korea climbed from #88 to #18, Japan from #96 to #21, Taiwan from #60 to #19, Hong Kong #49, and the US bounced between #160 and #200. It never charted on any country's Top Grossing.

Discount these positions. The Korean Entertainment free chart #18 slot had Dearbe at 11-10 with 20.8k cumulative installs, while #20 was Crack with 1.14M installs. The Korean and Japanese entertainment free charts are crowded with short-drama apps, and two or three thousand new installs a day is enough to reach the top 25. Only 0.22% of its 46 installs left a rating, averaging 3.27, which is typical of bought users: they installed, barely used it, and didn't bother to review.

### Where users come from

A game company's launch playbook, moved onto an AI chat app:

- **Same-day press releases** : on 11-02, Korea's ZDNet and GameMeca and Taiwan's Bahamut published launch news at the same time; the next day Game Chosun and Ruliweb followed with nearly identical copy, every image credited "Source: 디어비." All of the outlets were gaming media.

- **Contests plus token giveaways** : a ₩15M creator contest (11-02 to 11-22, open to Korea, Japan, Taiwan, Hong Kong, and the US), 11-30 Spark for signing up before 1,000, another 2,000 after one more verification step; a dedicated redemption code DCAIINFO posted on the Korean forum dcinside for 500 extra. Each channel got its own code, so you can tell at a glance which source brought in the most people.

- **Poaching creators from rival platforms** : the Writer Onboarding Program (through Jan 30, 2027) is explicitly aimed at "creators already active on existing AI character chat platforms." They come in at the top tier right away, rewards are doubled, and they can cash out. 11-02 The most-liked Korean five-star review says "the illustrations I drew before can finally be used here." You can tell it's an old player moved over from another platform.

- **Paid influencers** : the Instagram post at the top of this report.

- **Paid acquisition** : both one-star Japanese reviews say they found the app through an ad. The official TikTok account @dearbe_official has only 3 followers but has posted 145 videos, about one an hour, each getting two to six organic views. That's a creative library for feed ads (inference, high confidence). Every official link runs through AppsFlyer attribution.

We found zero organic word of mouth: nobody mentions it on X, TikTok search turns up no user-made content, and the only post in dcinside's AI chat gallery is the official one. The official Instagram has 86 followers.

### What users say

There are few reviews, but they are telling.

The earliest Korean reviews are positive, from 11-01 to 11-03, with ten to twenty likes each: "The interaction (tapping the character) is fun, I want to try making one myself" and "Different from other chat apps." The timing lines up with launch and beta, most likely from testers and creator circles (inference).

One specific four-star US review from 11-04 reads: "Please add a free version that uses no Spark and lets you watch ads to chat. Right now I have to wait for Spark to recover before I can continue, and it's really annoying." A two-star review from 11-05 lists it out: "Scam: ⚠️ Personas: ✅ No filter: ❌ Tokens / paywall: ⚠️ Ads: ⚠️ Pushes payment: ⚠️ Works on web: ✅ AI smarts: ⚠️ Unlimited: ❌"

The Japanese store is the two reviews at the top: not what the ads showed, text too small, not as good as zeta.

In short, what draws people in is the derivative mechanics (tap interactions, photos together, songs). What turns them off is tight token gating and chat quality that trails the established apps.

### Who it's competing with

AI character chat is one of the most crowded categories in Korea and Japan:

| Product | Scale | Versus Dearbe |
|---|---|---|
| zeta(Scatter Lab) | Google Play 7.84M installs, +7k over the last 190 days, best Top Grossing rank #11 in Japan; MAU 1.5M, revenue around ₩2025B in 26 and profitable (Korean press) | Korea and Japan leader, about 300× Dearbe's scale; Japanese users compare Dearbe directly to it |
| Crack(Wrtn) | 1.15M installs, best Top Grossing rank #11 in Korea; parent company valued at about ₩1000M | Web-novel-style chat with creator revenue share for a long time. The writers Dearbe is poaching come from this group |
| BabeChat, Caveduck | 170K and 260K installs, Top Grossing around #100 in Korea | Mid-tier in Korea, chat plus images |
| チャチャ | 190K installs, Entertainment Free #26 in Japan | Local Japanese oshi character chat |
| Kiff(Krafton) | Launched the same day as Dearbe, monthly fee ₩3,900 | Big company, K-pop idols as official characters, not UGC |
| Talkie, Character.AI | 60.8M and 70.8M installs | Global giants; Talkie already has character card collecting and voice |

(Install counts are Google Play, early November.)

The extra layer Dearbe adds (songs, photos together, figurines, fan shots) is not hard to build on its own: zeta and Crack both have image generation, Talkie has card collecting, and songs can come from a music generation API. What it really bets on is poaching creators from rivals and deepening the character library.

### Why this one

- **The money and people are in place.** Same-day game-media press releases, contests, tokens, channel codes, influencers, asset accounts, and attribution are the standard playbook of a mature publishing team. Behind it is a listed game company providing infrastructure, a setup one person or a small team can't get.

- **It picked low-barrier charts.** In Korea and Japan, Entertainment Free charts can be entered with a couple thousand installs a day, and reaching 2.52M installs earns a screenshot reading "#18 in Korea."

- **The ad hooks are well chosen.** "Take wedding photos with an AI character" and "Turn your conversations into a song" are things people can watch and share, far easier to advertise than "yet another smarter AI chat."

- **But it hasn't been proven yet.** No country has hit a top grossing chart, the rating is 3.27, and Japanese users say it's "not what the ads showed." Whether installs hold up after the 11-22 contest ends and the 11-30 token giveaway closes, and how the iOS launch (said to be mid-November) performs, we can't tell yet.

### If it were you

**One-line verdict:** there are a few tactics to learn, but don't go head-to-head. This is a late entrant backed by a game company, pushing into a category already crowded by zeta, Crack, and Talkie, relying on publishing budget, and it hasn't yet proven it can monetize. A small team building an "AI character chat platform with derivative mechanics" would need to solve content supply, model costs, and paid acquisition at once, and it would be weaker than Dearbe on all three.

**Who it fits:** teams already running AI character chat that have a user base and creators, who can take Dearbe's tactics to boost their own numbers: use "chat → photos / songs" as ad creative, run small contests and channel-specific codes at launch, and target rivals' creators directly.

**Who it doesn't fit:** individuals and small teams trying to build an AI companionship platform from scratch. The leaders in Korea and Japan are already profitable, and a late entrant without budget can't get in the door.

**How to do it:** if a small team needs an opening, pull out the most profitable piece of Dearbe on its own: don't build a chat platform, build a small web tool where existing zeta, Crack, and Character.AI users upload their character settings or chat screenshots and get back a four-panel photo with their character, a wedding photo, or a personal song, charged per image or per song (inference, not yet validated whether anyone will pay). The chat belongs to someone else. You sell the "merch," you don't compete with the platforms for users, and you survive on their users.

**How to get users:** post finished images where zeta and Crack users gather (the AI chat galleries on dcinside, the oshi chat circles on Japanese X). Watermark each photo with a link so every shared image becomes an ad. Copy Dearbe's move: give each channel its own redemption code so you know which source brings paying users.

**When to decide, and before when:** before late January 2027. Dearbe's Writer Onboarding Program ends Jan 30. By then it will be clearer whether it keeps its creators and whether derivative mechanics drive payment. zeta and Crack could also add photos and songs at any time.

**Signs that would mean you read it wrong:**

1. After the 11-30 token giveaway ends, if Dearbe's daily installs stay above two or three thousand, or it makes any top grossing chart in Korea or Japan, that shows the derivative mechanics really retain users and earn money. Then the call is worth revisiting, and the rating goes up.

2. If zeta or Crack launches its own photo or song feature, the "merch" angle closes.

3. If Dearbe's rating drops below 3 after December and installs stall at thirty or forty thousand, that means it was a bought launch and the derivative mechanics are just ad creative. Even the lessons to learn lose value.

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

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

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| zeta(Scatter Lab) | Android | 2022-05 | Play 7.84M installs, +7k over the last 190 days; best Top Grossing rank in Japan #11; MAU 1.5M, revenue around ₩2025B in 26 and profitable | Korea and Japan leader, pure chat plus images. Dearbe adds songs, four-panel photos, figurines and other derivative content, but at about 300× the scale. Japanese reviewers say zeta is more fun. |
| Crack(Wrtn) | Android | 2025-03 | Play 1.15M installs, best Top Grossing rank #11 in Korea; Wrtn valued at ₩1000M, target annual revenue ₩100B+ | Web-novel-style chat with creator revenue share for a long time. Dearbe's Writer Onboarding Program targets creators on exactly this kind of platform. |
| BabeChat | Android | 2025-02 | Play 170K installs, Top Grossing #122 in Korea, #2 in Korean search volume | Chat plus image generation, no collecting mechanics like songs, video or figurines |
| Caveduck | Android | 2025-01 | Play 260K installs, Top Grossing #99 in Korea | Mid-tier pure character chat in Korea |
| Kiff(Krafton) | Big-company features | 2026-11-02 | Launched the same day, FIFTY FIFTY involved, monthly fee ₩3,900 | Big-company official K-pop idol characters, not UGC |
| チャチャ(Kinkaku) | Android | 2025-09 | Play 190K installs, Entertainment Free #26 in Japan | Local Japanese oshi character chat |
| Talkie | Android | 2023-06 | 60.8M Play installs | Global giant that already has character card collecting and voice, partly covering the "beyond chat" mechanics |
| Character.AI | Android | 2023-05 | 70.8M Play installs | The largest global UGC character chat, with few derivative features |

Where users come from

- **Verified** **press release (game media)** : 11-02 On the same day, ZDNet Korea, GameMeca, and Taiwan's Bahamut GNN all ran it. 11-03 Game Chosun and Ruliweb followed, with images credited as 'Source: 디어비'

- **Confirmed** **Launch creator contest** : 11-02–11-22, total prize pool ₩15M, across five regions (Korea, Japan, Taiwan, Hong Kong, US); contest videos on YouTube and Instagram

- **Confirmed** **Signup tokens + channel-specific codes** : 11-30 signups before 1,000 Spark; "verify Unlimited" adds 2,000; dcinside code DCAIINFO gives 500

- **Confirmed** **Poaching creators from rivals** : the Writer Onboarding Program targets "creators already active on existing AI character chat platforms": direct upgrade to Diamond, rewards doubled before 1-30, cash-out available (official dcinside post)

- **Confirmed** **Paid Instagram influencers** : AI creator @prompt_what posts #광고 content focused on wedding photos and four-panel shots with characters, using comment replies to send DM links

- Inference **TikTok / YouTube ad creative accounts** : TikTok @dearbe_official has 3 followers but 145 videos posted about hourly, each getting 2–6 organic views; YouTube has 24 subscribers and 25 videos. Our ad library does not track it.

- **Confirmed** **Paid acquisition (Japan)** : 1 Japanese one-star reviews: "I saw this app in an ad" and "totally different from the ad content" (Play reviews 11-04 and 11-07); every official link runs through AppsFlyer OneLink.

## Dearbe: More Than AI Chat (com.dearbe.dearbe) research report

Research date 2026-11-09 (data cards refreshed to 11-10). Sources: googleplay / appstore / metaads / tiktokads / googleads libraries (agent-im server), Jina web search and full-page scraping, TikHub (TikTok), x-radar, public Google Play reviews (live-scraped with google-play-scraper).

### Bottom line

Dearbe (Korean name 디어비) is an AI character companion + UGC platform from Seoul's DEARBE Corp. You chat with AI characters, and you can also turn chats into songs (OSTs), illustrated novels, couple photos ("four-cut" / wedding photos) with characters, fancam videos, and virtual figurines. 11-02 launched on Google Play and web at the same time, selling in KR / JP / TW / HK / US. The iOS release is said to land in mid-November.

**It hasn't really taken off yet.** About 21k installs in the first 9 days: a well-prepared paid launch. Korean and Taiwanese game press releases, a ₩15M (about $10k) creation contest, 1,000–3,000 Spark tokens for signing up, poaching creators from other AI character chat platforms (with doubled rewards), paid Instagram influencer promos, plus TikTok / YouTube ad creative accounts and AppsFlyer attribution. I can't find any organic spread. Its Korean and Japanese Entertainment Free charts hit #18–#24, because those category charts have a low bar: one install a day every 2,000 is enough. It hasn't charted on any store's Top Grossing. The backing likely comes from listed game company Joycity (조이시티): the privacy policy says Joycity runs its cloud infrastructure, login platform, and payment integration.

### 1. Who they are

- **Company** : DEARBE Corp. (주식회사 디어비), address at 麻浦大路 92, Mapo District, Seoul [dearbe.me/legal/terms]. The representative and personal information officer is Bae Jun-seok (배준석) [dearbe.me/legal/privacy; ZDNet Korea 2026-11-02].

- **Link to Joycity (strong association, ownership not verified)** :
    - The "entrusted processing" section of the privacy policy says **Joycity Corp. handles "cloud IT infrastructure operation / management, member registration and login platform, and payment service integration development"** . The third-party behavior data section lists Joycity's "Common Log" SDK [dearbe.me/legal/privacy]. When a startup hands core systems like accounts and payments to a listed game company (KOSDAQ 067000, developer of Freestyle and Gunship Battle), it usually means a subsidiary, an incubated project, or a deep partnership.
    - In 2014 coverage from ZDNet and Game Donga, Joycity's "Publishing Division Director Bae" spoke publicly [zdnet.co.kr/view/?no=20140430180301; game.donga.com/72990]. Same name as Dearbe's representative, possibly the same person, **not verified** . (Another person with the same name is the representative of game outsourcing firm GameDex, based in Seongnam, which doesn't match Mapo, so probably not him.)
    - The launch playbook is the game-publisher playbook: game press releases (ZDNet's game desk, GameMeca, Game Chosun, Ruliweb, and Taiwan's Bahamut GNN, all the same day), event-style launch activity, token giveaways, and promo codes.

- **Tech stack** (per privacy policy): OpenAI, Anthropic, and Google models; Tencent Cloud hosting DeepSeek and video generation; Inworld AI for voice calls; Microsoft PhotoDNA / Azure Content Safety for image moderation; TypeSafe AI for chat classification; AppsFlyer + Google for ad attribution; AdMob for ad monetization; Danal for Korean local payments.

- The website's CDN path is called `pocketmaker`. The internal project codename may have once been PocketMaker (speculation).

### 2. The product

Sources: Play store listing, DCInside official promo posts, Korean press releases.

- Chat comes in two styles: message-stream, or "novel-style." 35+ genres (romance, BL / GL, regression, reincarnation, wuxia, idols, chaebols...), clearly built for Korean web-novel readers.

- **Content beyond chat, and this is its main selling point** :
    - OST: turn conversations into songs, in vocal and instrumental versions;
    - My Story: turn conversations into illustrated novels;
    - Couple photos: generate four-cut photos, wedding photos, magazine covers, and movie posters from "one photo of mine + the character";
    - Video hall: character fancams and challenge dance videos;
    - Figure Room: collect virtual figurines of characters and see what the figure "talked about last night";
    - Memory cards, character social feeds, and real-time voice calls;
    - "Action": tap different parts of a character's body, and expressions and reactions change in real time.

- **Creator side** : generate a character draft from a one-line concept; lorebooks, status bars, affection meters; anime / semi-realistic / realistic art styles plus animated GIFs. Becoming an "official creator" earns revenue share.

- **Monetization** : Spark tokens. Free ones come from daily check-ins and tasks; the rest is in-app purchases, $0.79–$77.90 per transaction [googleplay library app_snap]; plus AdMob ads. A DCInside post mentions an "Unlimited" tier.

- Content rating: 12+ on Play; the contest is limited to adults.

### 3. Growth data (googleplay library)

| Date | Total installs (Play maxInstalls) | Score | Average | Version |
|---|---|---|---|---|
| 11-04 | 6 (count not updated) | – | – | 1.2 |
| 11-05 | 3,812 | 16 | 3.17 | 1.2 |
| 11-06 | 3,812 | 24 | 3.29 | 1.3 |
| 11-07 | 3,812 | 31 | 3.00 | 1.3 |
| 11-08 | 15,182 | 37 | 3.00 | 1.3 |
| 11-09 | 16,580 | 42 | 3.17 | 1.5 |
| 11-10 | 20,812 | 46 | 3.27 | 1.5 |

- The count is still 11-04 6, but the same day it was already #88 on the Korea Entertainment Top Free chart and #60 in Taiwan, so the Play count is clearly lagging. Sitting at 3,812 for several days and then jumping to 15,182 is also just how the count updates, not a real 11k gain on any single day. **Counting from the 11-02 launch, that works out to about 2,300 installs a day on average.** AppstoreSpy hasn't backfilled its history for it (app_day_ext is empty).

- **Charts** (chart_rank, all Entertainment Free charts):
    - Korea 88 → 43 → 21 → 18 → 18 → 19 → 18;
    - Japan 96 → 66 → 38 → 26 → 21 → 21 → 24;
    - Taiwan 60 → 35 → 19 → 19 → 20 → 21 → 23 (best overall App Store rank in Taiwan: #149);
    - Hong Kong 125 → 49;
    - US: from 11-06, hovering between 163 and 198.

- **Low bar** : 11-10 Dearbe, #18 on Korea's Entertainment Free chart, has only 20.8k total installs, while the #20 right beside it is Crack with 1.14M installs. In Japan, Dearbe at #24 ranks above チャチャ (#26) with 190k installs. These free category charts are packed with short-drama apps, and two or three thousand new installs a day is enough to reach the top 25.

- **No Top Grossing chart placement at all** (googleplay library best_grossing_rank is empty).

- 46 ratings, only 0.22% of installs, average 3.27, a low engagement ratio typical of paid-acquisition traffic.

### 4. Where users came from (growth tactics)

| Channel | Evidence | Confidence |
|---|---|---|
| Press releases | On 11-02, ZDNet Korea, GameMeca, Daum (reposted), and Taiwan's Bahamut GNN ran the same story; 11-03 Game Chosun and Ruliweb followed with nearly identical copy, images credited "Source: 디어비." Launched the same day as Krafton's Kiff, same reporter. | Confirmed |
| Launch creation contest | 11-02–11-22, ₩15M total prize pool, open to all five markets; contest videos on YouTube and Instagram [gamemeca; YouTube -Q00HqDPOrA]. | Confirmed |
| Token incentives | Before 11-30, signing up gave 1,000 Spark; "verifying Unlimited" added 2,000; DCInside exclusive code DCAIINFO gave 500 [dcinside aichatinfo/329]. | Confirmed |
| Poaching creators from other platforms | "Writer onboarding event" (11-02–1-30): aimed at "creators active on existing AI character chat platforms," given the top "Diamond" tier directly, with doubled rewards for the first 1-30 people, and withdrawable. Early Korean reviews include someone saying "the illustrations I made before can now be used" [dcinside; Play review 11-02]. | Confirmed |
| Paid Instagram influencers | AI creator @prompt_what (프롬왓) posted an #광고 (ad) post: wedding photos and four-cut photos with characters, promoting the contest and token gifts, with a "comment '디어비' and I'll DM you the link" hook [instagram.com/p/DfZaLOAFO_V]. | Confirmed |
| TikTok / YouTube ad creative accounts | TikTok @dearbe_official has only 3 followers but has posted 145 videos, about one an hour, each getting organic views of 2–6. It looks like a creative pool for Spark Ads. The official YouTube channel has 24 subscribers and 25 videos [TikHub; YouTube]. It doesn't appear in the European TikTok ad library or the Meta / Google libraries (it's not on our tracking list, so not finding it there doesn't mean it isn't advertising). | Strong speculation |
| Paid ads (Japan) | Two Japanese 1-star reviews: "Saw this app in an ad" and "Nothing like the ad, zeta is more fun" [Play reviews 11-04, 11-07] | Confirmed (ads running; platform unclear) |
| Attribution infrastructure | The official Instagram / TikTok links are all AppsFlyer OneLinks (dearbe.onelink.me); the privacy policy names AppsFlyer and Google ad measurement. | Confirmed |
| Web | Launched on PC web (dearbe.me) at the same time as Play; creator posts say "you can play right on the web too." | Confirmed (share of traffic unknown) |
| Organic community buzz | X (x-radar): 0 posts; searching TikTok for dearbe / 디어비 returns nothing relevant; DCInside's AI chat board has only one official post (121 views); official Instagram has 86 followers. | None |

**Where the money comes back** : Spark token in-app purchases (top tier $77.90) plus AdMob ads, the same model as zeta / Crack. But it has no Top Grossing placement so far, so paying users are unproven.

### 5. Competitors

| Competitors | Platform | Launched | Scale | How it differs from Dearbe |
|---|---|---|---|---|
| zeta(제타,Scatter Lab) | Android / iOS | 2022-05(Play) | Play: 7.84M installs, +7k in the last 190 days; best Japan Top Grossing rank 11, Japan Entertainment Free #2; MAU 1.5M (2025-12), ~₩2025B revenue in 26 and profitable [googleplay library; etnews 2026-01-14; sedaily]. | The leader in Korean and Japanese AI character chat, pure chat plus images, with a scale several times that of Dearbe 300; Japanese user reviews compare it directly |
| Crack(크랙,Wrtn) | Android / iOS / web | 2025-03(Play) | 1.15M installs on Play. Best Korean top-grossing rank: #11. Wrtn's valuation: 10k hundred million won. Target annual revenue: ₩100B+, with more than half from Japan and the US (googleplay library; Chosun Ilbo 2026-08-26; Hankyung 2026-10-18). | Web-novel-style AI chat that already pays creators a revenue share ("earn from AI characters like a YouTuber"); Dearbe's "writer onboarding" is aimed squarely at its creators |
| BabeChat(바베챗) | Android / web | 2025-02 | Play 170K installs, Top Grossing #122 in Korea | Korea's #2 character chat by search volume, with image generation |
| Caveduck(케이브덕) | Android | 2025-01 | Play 260K installs, Top Grossing #99 in Korea | Mid-tier character chat in Korea |
| Lofan AI(로판AI) | Web / App | 2024 | #3 in Korea by search volume (160k searches / month) [Facebook post citing search volume stats] | Romance-genre vertical |
| Kiff(키프,Krafton) | App | 2026-11-02 (same day) | Newly launched, with FIFTY FIFTY involved; monthly fee ₩3,900 [Yonhap 2026-11-02] | Big-company official K-pop idol characters, not UGC |
| チャチャ(Kinkaku) | Android | 2025-09 | Play 190K installs, Entertainment Free #26 in Japan | Local Japanese oshi character chat |
| Talkie(SUBSUP / MiniMax) | Android / iOS | 2023-06 | 60.8M Play installs | Global giant, already has character card collecting and voice, and some of the "beyond chat" gameplay too |
| Character.AI | Android / iOS / web | 2023-05 | 70.8M Play installs | Largest globally, UGC characters; little derivative content |
| Saylo / Linky / Emochi, etc. | Android / iOS | 2023–2024 | Play 8.45M / 23.8M / 30.1M installs | Export-focused character chat that leans heavily on paid acquisition |

What I could find: this track is extremely crowded, especially in Korea and Japan. Dearbe is a late entrant, and its scale is still two to three orders of magnitude behind.

### 6. What's the differentiation, and is it the reason it took off?

- **Differentiation** : treat "chat" as raw material and turn it into collectibles and shareables: OST songs, illustrated novels, couple four-panel photos / wedding photos using the user's own face, fan-cam videos, virtual figurines, tap reactions. The product positioning is a three-stage "reader → creator → collector" funnel, layering "goods" (merch) consumption on top of text-only experiences like zeta / Crack. It borrows the "life four-cut", "fan cam" and "OST" formats Korean users already know.

- **Is that what made it take off?** : **Not yet.** Almost all of these 20k installs came from distribution pushes (press releases, contests, free coins, influencers, paid ads). The differentiation so far mainly gives the ads creative material: an influencer post pitches "take wedding photos with AI characters," which sells better than "yet another AI chat." But Japanese users say "totally different from the ads" and "text is too small, uninstalling right away," with a rating of 3.27, which shows the promise in the creatives hasn't turned into retention yet. The earliest 5 one-star reviews in Korea (11-01 to 11-03, 14 to 20 likes) most likely came from beta testers or the creator community.

- These features are not hard to copy. Talkie has cards, zeta / Crack both have image generation, and Suno-type APIs are available to anyone. The moat depends on whether it can poach creators from Crack / zeta and thicken the UGC library.

### 7. How many people have seen it

- Media coverage is limited to Korean IT / gaming outlets and launch press releases on Taiwan's Gamer, with no in-depth reporting and no English coverage. There is 0 discussion on X, no organic content on TikTok, and nothing found on Reddit.

- Revenue is not public, and there is no best-seller ranking.

- **Verdict: almost nobody knows about it yet.** Everything visible is content it paid to put out.

### 8. What to borrow

1. **Use "chat → shareable derivatives" as the paid-acquisition hook** : visual, shareable outputs like wedding photos, four-panel photos, and songs written for you beat "a smarter AI" by a wide margin for paid media, and suit influencer and feed ad creative.

2. **Cold-start by poaching competitors' creators** : UGC platforms are most starved for content. Offering "creators active on other apps" the top tier, double revenue share, and a time-limited window is faster than growing creators from scratch.

3. **Contest + coin giveaways + channel-specific codes** : use a modest prize pool (about $10k) for a "global contest" as a news hook and content-sourcing tool, then issue a unique code per channel (DCAIINFO) to track conversion by channel.

4. **Porting the game publishing playbook to AI apps** : same-day press releases in multiple regions, climbing low-barrier category charts, OneLink attribution, and batch video posting from creator accounts. Publishing teams from listed game companies running AI companion apps are essentially transplanting this playbook.

5. **"Launch on Play + web first, iOS two weeks later"** : Play and web review are fast, so creative and payment tests can run first, with iOS going live once data is in.

### 9. Still unclear

- What the relationship is between DEARBE Corp. and Joycity (subsidiary / incubation / outsourcing), whether 배준석 was the former Joycity publishing head; funding details were also not found.

- How much was spent on paid acquisition and on which platforms (Meta / TikTok / Google are not on our ad-tracking list).

- Retention and payments: no best-seller ranking and too few reviews; it will be clearer after the iOS launch (expected mid-November).

- Whether installs can hold at two to three thousand a day after the 11-22 contest and 11-30 coin giveaway end.

- Traffic share of the web version dearbe.me is unknown (no rank if it's outside Tranco's top 1M; I didn't check).

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