15k installs, self-reported $1M annual revenue: Plaitoon, a women-focused AI chat app built by 3 people in Korea

Plaitoon is an AI character chat built by a Seoul company of 3 people, focused only on women's content: romance, romantic fantasy and BL. It hasn't gone viral; Google Play shows only 15k cumulative installs, adding a hundred or two a day. But the team's hiring page says it broke even in 2026 February and reached $1M ARR in April. Its money comes not from acquisition but from getting a small group of heavy users to pay message by message.

A creator "moved" the characters

On Dec 2025 30, a Korean creator going by 「금가락지」 posted on Postype, the Korean writing platform, with the title 「플레이툰(여성향 전용)캐릭터 챗 이식」, meaning moving characters to Plaitoon (women-oriented only). She used to build characters on Crack, one of Korea's biggest AI chat platforms, and this time she brought over the full cast from two of her works.

Her reasoning is practical: 「The system and pricing are about the same as Crack, but they fixed Crack's weak points and absorbed the strengths of other apps.」 Then she listed a bunch: branching storylines (「time machine」), chat logs you can download, longer prompts, and a reading view where you can adjust font size and typeface like an e-book. She also tested output length: for the same character, Gemini 2.5 gave about 1,700 characters per message for 80 gems, and Gemini 3.0 gave about 2,100 characters for 100 gems. Crack's default settings gave 1,200 and 1,700 characters. One more thing excited her: 「Claude Sonnet 3.7, which was retired elsewhere, is still alive here.」

The post ended with her referral code: 「Enter my friend code and get 777 jelly.」

This one post is almost the whole Plaitoon playbook: give heavy users longer, more controllable stories, let creators move their characters and readers over with them, and use referral codes to pull people in.

What it is

An AI character chat that works like web fiction. The first line of its Korean App Store description reads 「In richly described, web-novel-style stories, keep going with your characters in a story that belongs only to you.」 Main features:

• Long-form scenes: not short back-and-forth messages, but paragraphs with expressions, environment, and inner monologue. A Japanese user wrote: 「Dialogue and narration are about half and half, just right.」 (Japan App Store, 9-18)
• Memory: important events and promises can be saved to memory, so the character still remembers when you come back days later. 「I played for free, came back to talk to it three days later, and it remembered. I was shocked.」 (Japan App Store, 10-09)
• Branching: go back to any point and take a different path.
• Choice of models: Claude, Gemini, plus a self-tuned model leaning adult. Pricier models write longer, cheaper ones write shorter.
• The character's 「phone」: you can browse the character's Instagram-style posts and search history, and characters message you first or call in the morning to wake you up. Reviews praise this feature the most: 「You can snoop on the character's phone!!!!」 「Every time I scroll through that Instagram and search history, I can picture the character's private life. That's the Plaitoon flavor.」 (Japan App Store, 10-09, 10-11)

The web version launched in public beta in June 2025, the iOS app went live in September 2025, and the Android app came much later, with the earliest reviews in April 2026. It now has Korean, Japanese, and Traditional Chinese interfaces.

How big is it really?

Two things need to be read together. First, it is really small: Play only added 1,200 installs in a week, and iOS ratings barely moved on either store that week, so new users are few and revenue mainly comes from existing users spending steadily. Second, it sits on the bestseller chart because it picked the 「Books」 category. Most similar apps are listed under 「Entertainment」, where you'd need hundreds of thousands to millions of installs to crack the top 20. The Books top chart is led by web-novel platforms, so the bar is much lower; with 15k installs it can rank #16 (inference).

There's a strange detail too: of 1,894 ratings on Play, 1,558 are five-star and 222 are one-star, with no three- or four-star ratings at all, and only a dozen or so include text. The number of ratings is 12% of installs, while normal apps usually see 1–3%. It may be from an early campaign that rewarded ratings, or it may be an oddity in the store data. Not verified.

The $1M ARR figure only appears on the job posting; no media has reported it, and there is almost no discussion on X. At 15k installs, it would require a small group of users to spend around a hundred dollars a year each. That isn't outrageous for AI romance chat, but it is self-reported, so treat it as a ceiling.

Where the money comes from

Pay per message, plus a membership. Each message costs gems: Claude about 50 per message, Gemini 80–100 (tested by a Postype creator, 2025-12). Gems can be bought individually, with in-app purchases on Play from $0.79 to $89.99; there is also a monthly membership. Daily check-ins give 300 jelly (free currency), tasks grant trial membership, and a Halloween top-up returned 25%. Morning wake-up calls and similar features cost extra gems, 1,500 per use.

User comments show best how well this design works:

「Before I knew it, I was paying.」 (Korea App Store, 10-10)

「I chatted all day and the jelly melted away bit by bit. In the end I got a membership anyway.」 (Korea App Store, 10-13)

「How scary how fast the time and gems melt away.」 (Japan App Store, 10-01)

Some complain about the cost: 「It's more fun than other platforms so I keep using it, but I spend way too much money... please release a monthly unlimited chat card.」 (Korea App Store, 10-12) The price is real, but users who grumble about cost keep paying anyway, and that is the foundation of this business.

Why users are willing to spend this much: AI character chat in Korea has been booming for two years. Industry blogs estimate that the mass-market zeta had revenue of ₩2025B in 26.7; wrtn, which runs Crack, is valued at ₩10 hundred million; and the women-oriented Caveduck raised a Series A in April 2026. Heavy women-oriented users are longtime readers of web novels, otome games, and BL, and they're already used to paying per chapter for stories they love.

Where users come from

• Creator migration + referral codes: the post at the top is not the only one of its kind. The company says it has 70 active creators; referral codes give both sides 777 coins, and users even paste their codes directly into App Store reviews (Korea App Store, 10-04, 10-10).
• Vertical community word of mouth: the 「AI chat women-oriented」 board on Korea's dcinside, Naver blogs, and Postype all have long reviews and side-by-side comparisons. In Japan, an X player who tests AI chat apps wrote a note recommending it, saying they want to 「bring more Japanese users to Plaitoon, especially the dream-girl crowd」 (2026-01). A Japanese review says 「X's timeline pulled me in.」
• Light Google ads: the Google Ads Transparency Center shows about 30 ad creatives from the developer NETTY, the earliest from 2025-09. In the last week they ran 21, targeting Korea, Japan, and Taiwan, mostly search and image ads, at small volume.
• Official accounts: the Korean X account has posted 900 times or more, with one or two thousand views per post; a separate Japanese account exists, and Instagram features a character every day. TikTok has almost no presence.
• Things they barely do: no large-scale paid acquisition and no paid influencer partnerships spotted.

Who it's competing with

Lots. Korea's version of this business probably has 30 players, and more than a dozen focus only on women-oriented content.

A Korean user wrote a review of 20 women-oriented platforms on dcinside, ranking Plaitoon #14 with the verdict 「well-tuned in-house model, strong adult content, lots of features, but the basics are average.」 It isn't the best and it isn't the biggest; it survives on feature density and update speed, going from 1.2.27 to 1.2.31 updates in one week on Google Play.

The team is part of its edge too: 3 people, all developers, and the recruiting page says six days a week, 12 hours a day, 「using Claude Code and Codex as agent teammates」, and the company slogan is 「print money with AI.」 A small team means it can break even even with modest revenue.

If it were you

A verdict: demand and payment have both been proven, and the model isn't hard to copy, but in Korea the space is already crowded. What's worth building is taking the 「web-novel-style long-form + memory + pay-per-message」 package into a language where no one has built a women-oriented product yet.

Who it's for: teams or solo founders with someone who is a heavy reader of web novels, otome games, or BL and understands what that crowd wants; one or two people who can tune prompts and models themselves and keep per-message cost under control.

Who it's not for: anyone who wants a mass-market companion chat (that turf belongs to Character.AI and Talkie); anyone who can't handle content moderation and payment risk (adult content is where this business earns the most, and also the main reason apps get pulled from stores and payment channels refuse them); anyone hoping to grow fast through paid acquisition.

How to do it:

1. Pick a language and an audience: romance and BL readers in Spanish, Portuguese, Indonesian, or Thai, or romantasy readers in English. First check whether that market already has localized women-oriented AI chat, and whether there are rivals like Crack that are big but have short outputs and strict moderation that you can compete against.
2. Launch on the web first. Web payments avoid store commissions and let you set your own content limits; launch the app later, and list it under 「Books」 when you do.
3. Copy the heavy users' checklist: long outputs, editable memory, branching, model tiers with pricing, and the character's 「phone.」 The last one is the cheapest and the most popular in reviews.
4. Charge per message with a membership on top, so model costs track usage directly. Don't offer unlimited chat.

How to acquire users: first find creators in that language who already make character cards or fan fiction on other platforms, ask them to move their characters over, and offer referral codes and revenue share; then post long-form reviews in the relevant communities (Wattpad comment sections, Discord, fan subreddits, local forums). Use ads only to defend search terms.

When to move: within about a year. The Korean players are already expanding into Japan and Taiwan, LoveyDovey is already big in Taiwan and Japan, and sub-brands from large companies targeting women are coming out too. Once they finish multilingual support, 「longer outputs」 alone won't be enough of a selling point.

Signs you got it wrong: ① three months after launch, your weekly retention is far below the 50% Plaitoon self-reports, meaning readers in this language don't have Korea's habit of paying per message; ② model costs per paying user eat more than half your revenue, meaning pricing isn't passing costs on; ③ Plaitoon's fundraising in the second half of the year falls through, or its updates visibly slow down, meaning even its own $1M isn't stable.

Data and sources

• Installs, ratings, charts: Google Play and App Store charts by country, 10-28 to 11-06.
• Creator porting, model pricing and output length: Postype post by 「금가락지」, 2025-12-30.
• User quotes: App Store reviews from Korea and Japan, 2026-09 to 10, translated.
• Self-reported results, team, and way of working: NettY's company page on groupby.kr; job posting (2026-05).
• Market size: royvalue's Naver blog article on the Korean AI chat industry; Caveduck Series A from Korean media (2026-04).
• Platform ranking: dcinside "AI chat, women-oriented" board, reviews of 20 platforms.
• Ads: Google Ads Transparency Center, 11-06.
• ARR and breakeven are company claims with no third-party evidence; the split of revenue across web, iOS and Play is unknown; the explanations "low bar in the Books category" and the rating distribution are inferences.