# Feed screenshots of the person you like into AI to practice your replies: Koisimyu reached Japan a month late · Mobile app

> Source: https://stringers.io/en/apps/ios_6793537329.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 hasn't really taken off. Japanese young people are already hooked on the habit of feeding LINE chat logs into an app for analysis, driven by めろとーく and 恋愛温度計. It arrived a month later, rode a wave of LINE mini-drama Reels posted to its own Instagram account, and added hard paywalls, which got it onto a low-bar Lifestyle Top Grossing chart.

[Mobile App](https://stringers.io/en/apps/index.md) · iOS · written 11-17

こいシミュ · Ibuki Hara

# Feed screenshots of the person you like into AI to practice your replies: Koisimyu reached Japan a month late

**What it is** A Japanese dating chat practice app. Upload screenshots of chats with someone you like, and the AI mimics their tone so you can practice. Before you send something, it rewrites your message in three directions: gentle, natural, and one step further. Weekly subscription ¥600, plus pay-per-use.

**Why it went viral** It didn't really go viral. Young Japanese people were already in the habit of "feeding LINE logs to an app for analysis," a trend popularized by Meroto and Renai Ondokei. Koisimyu arrived a month later, and its own Instagram account's LINE mini-drama Reels brought in a wave of traffic. Hard paywalls then got it onto a low-bar lifestyle Top Grossing chart.

**C** **How you could follow** Don't build another one in Japan. If you speak Korean, Traditional Chinese, or a Southeast Asian language, port the "screenshot recreation + practice + polish before sending" loop there. Give generous free credits, add a shareable result page, and use short LINE-style drama videos to get users.

C grade: the product itself hasn't broken out (194 ratings, estimated monthly revenue in the low thousands of USD). Japan already has about fifty similar apps, and the leaders are an order of magnitude bigger, with ChatGPT underneath. What's worth copying is the short-video traffic template and the idea of switching markets. Whether other markets have room is unverified. Few people know about it yet

Est. monthly revenue (App Store) **$1K–10K** We estimated this from its chart ranks and rating growth across storefronts, calibrated against verified revenue; typical error is 2–3x. Read it as a tier, not an exact number.

Data (updated daily) · first listed 2026-07-29 · Japan rating 200 · charted in 11-20 of 1 stores

iOS · best category chart rank · by store

Data 2026-10-28 → 2026-11-20 (24 days, 18 observations). Lower rank is higher; gaps mean it wasn't charted that day

iOS · rating count · Japan

Data 2026-10-28 → 2026-11-20 (24 days, 18 observations). Apple doesn't publish download counts; rating growth is a proxy for downloads.

On Sep 19, a Japanese woman gave こいシミュ five stars on the App Store with the title 「修羅場になりました!」(it turned into a fight). She and her boyfriend were sharing a screen, cuddling while they played, and feeding in her ex's chat logs. "Why did it blow up, you get it, right? I'll probably be taken to court soon." The app does one thing: you feed in screenshots of your chats with someone, and the AI learns that person's tone so it can chat with you the same way. It was published by Japanese indie dev Ibuki Hara at the end of July, costs ¥600 a week, and now sits around #160 on the Japan Lifestyle Top Grossing chart. This piece explains what it is, where this amount of traction comes from, why we think it's late, and what's still left to build in this idea.

### What it is and how it charges

How it works in three steps: pick a few screenshots of your chats with the other person (LINE or Instagram DMs both work), the app reads the conversation, you check and tweak it, and it generates a "chat room." Inside the chat room you send messages, and the AI replies to you by drawing on how that person has written before: sentence length, tone, emoji. When you're about to send something for real, you can have the AI rewrite it in three directions: やさしく (gentle), 自然に (natural), and 一歩踏み込む (one step further). The chat room UI copies LINE, and the update notes for version 1.4 specifically mention "readjusted how closely LINE-style chats are reproduced." Later it added a "Love Notification Report" that scores your relationship like a report card.

The paywall is hard. The free tier covers about ten back-and-forths. Watching a video ad each day earns a little more credit, after which you have to pay. The top-selling in-app purchase is the weekly subscription at ¥600 (about $4), followed by 100 replies for ¥480, 10 chat-room tickets for ¥480, the monthly subscription at ¥1,980, and 300 replies for ¥1,200 (APPLION in-app purchase ranking, November).

Users mostly praise one thing: how much it sounds like the person. "Uncanny likeness!" (8-01), "It's so spot-on I can't stop chatting" (10-04), "It says things they'd actually say" (11-02). The complaints center on money: "Credits run out fast and then you can't keep talking" (9-21), "Practice shouldn't need a ticket" (11-02). Some also say it doesn't pick up how the person talks at all, and even gets their gender wrong (10-04).

### 194 ratings, chart rank 160, take with a grain of salt

From its launch in Jul 29 to Oct 28, it collected 166 ratings in three months, averaging under 2 a day. By Nov 18 it was at 194 ratings with a 4.2 average. On Nov 17 it gained 9 in a single day, while the other days ran 0 to 3 a day. I couldn't find out why that day jumped (App Store Japan storefront, 10-28 to 11-18). Only 13 ratings include written reviews. Most ratings came from an in-app prompt.

It has charted in only one place: Japan's lifestyle Top Grossing chart. From late October to mid November it bounced between #125 and #196. It never made the Top Free or overall charts. The bar for Japan's lifestyle Top Grossing chart at around #150 is low. Going by rank, monthly revenue is roughly in the $1,000 range, and at the high end maybe a bit over $10K (estimated). It's Japanese-only, with no Android version.

The developer has shipped 12 unrelated apps to the App Store since July: a walking check-in app, an English-Japanese dictionary, a few mini-games, and one more new one in Nov 17. Apart from こいシミュ, none has a rating above 30. This is someone batch-building and batch-testing apps with AI, and こいシミュ is the one that hit. It's a small win, though, not a breakout.

### Where users come from: a single Instagram account they run themselves

Sep 22, a two-star review explained where it came from: "I installed it because I saw on Instagram that it was good" (「インスタでいいと書いてあったので入れました」). Tracing that, nearly all the Instagram content I could find came from the app's own account, @koisimyu, which was originally called 「恋愛トーク練習」 (romance chat practice). I found no third-party creators or organic user posts.

This account started posting Reels on Aug 1. The content is LINE chat mini-dramas, like the Aug 4 post 「彼氏が夜中に送って、1分で消したメッセージ」 (a boyfriend sends a message late at night and unsends it a minute later). There are also tutorial-style videos that soften 「受かった!!」 ("I passed!!") into 「ちょっといいことあった」 ("something good happened"), and MBTI topics to drive engagement. Every post ends with the same line: 「物語に出てくるAIは、プロフィールのリンクから使えます(最初の結果まで無料)」: the AI from the story can be used via the profile link, free until the first result. No faces are shown, and it can be produced in bulk.

The approach wasn't invented by Koisimyu. A month earlier, Renai Ondokei started two persona accounts on TikTok in early July and mass-posted LINE conversation videos. By early August one single video had reached 30k views. Koisimyu just moved the same playbook to Instagram. The exact same caption also appears on Instagram for a web product called 仲裁くん (Chūsai-kun) in October. Whether it's the same person, the same agency, or copying, I don't know.

Searching Meta's ad library for its name and App ID returns 0 ads. Nobody discusses it on X, and no media outlet has written about it. The only results are auto-generated app listing pages.

Then the account stopped promoting it. @koisimyu is now renamed "walkstone - 歩いて磨く" and in November only posted about Walkstone, a walking app the developer launched in late October. With promotion stopped, ratings have mostly been growing by only one or two a day organically.

### It's a month late: who else is doing the same thing

By 2026, "send your LINE history to an app and see if they like you" was already an established habit among young Japanese people. Searching by name on the Japan App Store turns up 50 similar apps, most of them launched this year.

| Competitors | Launched | Scale | Compared with Koisimyu |
|---|---|---|---|
| Meroto (LINE compatibility analysis) | 2026-01 | 7,419 ratings on the Japan store, Productivity chart rank 69–88; the official line says the series has passed 1M cumulative downloads; estimated monthly revenue about $24K | The category leader: import exported LINE text and get a compatibility score card you can share to Stories. Later it also added "AI that imitates the other person's chat style." |
| Renai Ondokei | 2026-06-30 | 1,592 ratings on the Japan store, lifestyle chart rank 107–196; estimated monthly revenue about $4,600 | The most direct rival, a month earlier. It also has a "looks like the real person" AI persona, plus an extra feature: "if you'd said this instead, how would they reply?" Pricing is nearly identical (¥600 a week plus tickets). Some reviewers on Koisimyu's page say it's too expensive and switched to Renai Ondokei. |
| Clonely | 2026-11-03 | Just launched, on both the Japan and US stores | An AI persona app spun out from Renai Ondokei's developer, colliding head-on with Koisimyu's core feature. |
| タラレバDM | 2026-08-14 | 0 ratings | Nearly identical selling points (mimicking the other person's sentence endings, punctuation, and message rhythm). Launched two weeks later. It didn't take off. |
| Smond, Moteru Reply AI, AI Reply Navigator, and about ten more. | From ¥2025-08 | Mostly 0–3 ratings. | Upload a screenshot and get three reply suggestions right away. Koisimyu adds a step: it recreates the other person first, then lets you practice against them. |
| ChatGPT / Gemini | Already around | — | You can get the same result by uploading screenshots and asking the AI to play the other person. Koisimyu just packages the workflow in a LINE-style interface. |

It differs from the others in three ways: it builds the conversation from screenshots instead of exported LINE logs, which lowers the barrier and works with other chat apps; the interface is styled like a LINE chat room; and it offers three polished rewrite options before you send. But the core idea of "AI imitating the other person" was already in Meroto and Renai Ondokei, which are both an order of magnitude bigger. Screenshot input and polishing replies are standard features in many reply-suggestion apps. It got its current traction because the category was already hot and its own Reels brought in a wave of traffic. On the product side, it's "no worse than the others," not meaningfully ahead.

### A few moves worth noting

- **Story-style short video + "the AI from the story can be used via the profile link"** : LINE mini-dramas need no face on camera and scale easily, and the ending routes viewers to an AI tool that recreates the plot. This template has shown up with Renai Ondokei (TikTok), Koisimyu (Instagram), and 仲裁くん (Instagram), and it works for topics like romance and relationships. One product gets one dedicated account; when the product stops being promoted, the account is renamed and reused for the next product.

- **Ride a hot category with a side feature** : don't go head-to-head on the hottest compatibility scoring. Pick a side angle like "practice partner" or "polish before you send", build it solo, and ship it within a month. Charge the same weekly subscription plus per-use credits that already-proven apps use.

- **Lessons learned** : the free tier was too small, nearly every bad review was about paying, and a rival poached users. When the promo account was switched to pushing a different app, growth stopped. Mass-producing apps just means each one ends up shallow.

### If it were you

**C** . Worth learning from, not worth following up on. こいシミュ didn't break out on its own: 194 ratings and #160 on the Top Grossing chart, estimated monthly revenue in the low thousands of dollars. Japan already has around fifty similar apps. The leader is an order of magnitude bigger and keeps adding AI avatar features, with the free ChatGPT underneath. Launching another one in Japan means being the fifty-first.

- **Who it's for** : people who speak Korean, Traditional Chinese, or a Southeast Asian language, whose young users already chat about dating on KakaoTalk, LINE, or Messenger, and who can write realistic chat mini-dramas. Also people who can make vertical short videos and don't mind posting several a day.

- **Who it's not for** : anyone wanting to build yet another LINE analysis or AI avatar app in Japan. めろとーく, 恋愛温度計 and Clonely already split that market. If you can code but won't write scripts or make short videos, this kind of app has no technical moat, and everything depends on content accounts.

- **How to build it** : pick a different market and do the full loop: recreate the other person from screenshots, practice with them, then polish before sending. Screenshot input should not require exporting the chat, and the UI should mirror the messaging app. Give a generous free tier, and only charge once someone gets to the "it really feels like him" moment. Charge via weekly subscription plus per-use credits. Add a shareable result page (score card, report card). That share hook is what got めろとーく growing, and こいシミュ has nothing like it.

- **How to get users** : open a dedicated Instagram or TikTok account and post a daily chat mini-drama (a message unsent at midnight, read but no reply, an ex suddenly texting one line). End each with "the AI in the story is usable from the link in my profile, first result free." Post 30 before deciding, watch which storylines get saved and shared, then scale that direction.

- **Timing window** : before spring 2027. Young people in Korea and Taiwan have the same habit of screenshotting a chat and asking friends if someone likes them. I haven't seen anyone claim those markets (I haven't verified each one), but the Japanese leaders have started going to the US store, so it's only a matter of time before these markets get crowded. If daily new ratings don't reach 10 within 2 months of launch, stop pouring money into it.

- **What would prove me wrong** : by March 2027, if こいシミュ holds a steady top 100 on the Lifestyle Top Grossing chart without that Instagram account, or passes 1,000 ratings, then the practice angle keeps users and can grow, and there's room in Japan. Take it more seriously than I have. Conversely, if 恋愛温度計 and めろとーく have both dropped off the Top Grossing chart by then, the trend has cooled and there's no need to chase it in another market.

[iOS store page ↗](https://apps.apple.com/app/id6793537329)

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

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| めろとーく \| LINE相性分析 (appful) | iOS | 2026-01-18 | JP store 7,419 ratings, #69–#88 on the Productivity Top Grossing chart, the series claims over 1M downloads, estimated monthly revenue about $24K | Category leader. Import a LINE txt export to get a shareable compatibility score card; later added 擬人化チャット (AI imitates the other person's chat). こいシミュ only has practice, with no sharing hook. |
| 恋愛温度計 - もしもの恋愛シミュレーター (Issei Nishi) | iOS | 2026-06-30 | JP store 1,592 ratings, #107–#196 on the Lifestyle Top Grossing chart, estimated monthly revenue about $4.6K | The most direct rival, a month earlier. Also has an AI avatar, plus "もしも": rewrite a line from the past and see how the other person would reply. Pricing is nearly identical (weekly ¥600 + credits). Acquisition relies on 2 TikTok persona accounts, each posting 400+ videos. Some users said they switched from こいシミュ because it cost money. |
| Clonely(Issei Nishi) | iOS | 2026-11-03 | Just launched, on both the Japan and US stores | An AI avatar split into a standalone app, launched in both the Japan and US stores, head-on against こいシミュ's core feature. |
| IsTalk - トーク分析&相性診断 | iOS | 2021 | About 7,000 JP ratings | Long-running LINE analysis app with no generative AI that simulates the other person. |
| タラレバDM | iOS | 2026-08-14 | 0 ratings | You give it a few things the other person often says, and it replies to you in their style. Nearly the same selling point, didn't take off. |
| Reply suggestion apps (スモンド, モテ返信AI, AI返信ナビ, Relora, daptalk, etc.) | iOS | 2025-08 to 2026-09 | Mostly 0–3 ratings. | Screenshot in, 3 replies out. こいシミュ adds a step: recreate the other person first, then practice against them. |
| ChatGPT / Gemini, LINE AI Friends | Big-company features | Already there | Huge | Upload the screenshot to a general AI and ask it to play the other person and you can do the same thing. こいシミュ just packages the flow and wraps it in a LINE-style interface. |
| RIZZ (US store) | iOS | 2022 | iOS 68K ratings, estimated monthly revenue about $500K | English-market "AI helps you reply to texts". It offers reply suggestions, not a simulation of the other person, and doesn't target the Japanese market. |

Where users come from

- **Verified** **Own Instagram Reels account @koisimyu (formerly 「恋愛トーク練習」)** : from 8-01 onward posted Reels with #こいシミュ featuring LINE mini-dramas and tone-rewrite demos. Captions consistently said "the AI in the story can be used from the profile link (free until your first result)". Also posted MBTI engagement bait (8-18).

- **Verified** **Users saw it on Instagram and downloaded** : comments on 9-22 say "I downloaded it because the Instagram said it was good". I found no third-party creators or organic user posts, so this is most likely the official account's content.

- Inferred **Store traffic following similar-category heat** : 恋愛温度計 appears in the JP store's "customers also bought" section. The Lifestyle Top Grossing chart is a low bar.

- Inferred **Paid ads** : searching the Meta Ad Library by name, account, or App id returns 0 results each time, and the metaads database has no record.

## こいシミュ (iOS 6793537329) research report

> In one line: a Japanese indie developer, Ibuki Hara, built an AI romance chat practice app. Upload screenshots of chats with someone you like, and the AI learns their tone so you can practice replies. Before you send, it rewrites your draft in three directions: gentle, natural, or bolder. It never really took off. It's a late, small follower in Japan's "LINE chat analysis + AI avatar" wave. めろとーく (2026-01) and 恋愛温度計 (2026-06-30) had already made the category hot. It launched 7-29 later, promoted through its own Instagram account 「恋愛トーク練習」 (@koisimyu) with LINE mini-drama Reels. Weekly subscription ¥600 plus pay-per-use was enough to reach #125–#196 on the low-bar Japan Lifestyle Top Grossing chart. That Instagram account was later repurposed to promote the developer's new app Walkstone, and promotion for こいシミュ largely stopped.

### 1. How to read the data card (correcting first)

- **Very small scale** : Japan-store rating count went from 10-28 to 166 → 11-18 over 194,21 days, +28 (appstore database, app_day). In the 91 days since its 7-29 launch on 10-28, it collected 166 ratings, under 2 a day on average. In the recall, the "+16 over the last 7 days" figure included 11-17, which alone accounted for 9, while other days added 0–3 each. I didn't find why 11-17 jumped.

- **Top Grossing** : only charted in the Japan store's Lifestyle Top Grossing (genre 6012), bouncing between #125 and #196 from 10-28 to 11-18 (appstore database). Never hit the overall or Top Free chart. The bar for the Japan Lifestyle Top Grossing chart around #150 is low. Our database estimates monthly revenue around $1K, range $468K–$12K (stringer apps database, model estimate, use it only as a rough order of magnitude).

- **In-app purchases** (ranked list on the APPLION page): weekly subscription ¥600 ranks #1, followed by 100 reply credits at ¥480, 10 room-creation tickets at ¥480, monthly subscription ¥1,980, and 300 reply credits at ¥1,200. Most revenue comes from the weekly subscription plus per-use paid acquisition.

- **Ratings** : average 4.25 from 194 ratings. Distribution: 122 five-star, 25 four-star, 22 three-star, 10 two-star, 11 one-star (iTunes lookup and APPLION, 11-17). Only 13 reviews have written text (Apple JP review RSS), so most ratings were probably prompted in-app.

- **Japan only** : the interface and store page are entirely Japanese, and it only charted in the Japan store. The Google Play library has no entry, so there's no Android version.

- **Developer** : Ibuki Hara (developer id 6791958118). Since 2026-07 they've released 12 unrelated apps on the App Store (Life bar, momblock, Walkstone, ラストパイロット, ponkan English-Japanese dictionary, a few mini-games, and 「容量マジック」 added 11-17). Apart from こいシミュ, none of them has more than 30 ratings (iTunes lookup, 11-18). こいシミュ's support email is hearttrickjp@levipon.jp. levipon.jp is a Google review-management SaaS called レビポン (Supabase / Clerk / Vercel stack), so the developer is running other businesses too. Overall it looks like one person or a small team using AI to mass-produce and test products.

### 2. Competitors

Japan's "analyze or simulate your LINE chats with someone" niche has been very crowded since 2026. Searching the appstore library (app_name, jp) by name turns up 50 similar apps, most launched in 2026. The closest to こいシミュ:

| Competitors | Platform | Launched | Scale | How it differs from こいシミュ |  |
|---|---|---|---|---|---|
| めろとーく \ | LINE相性分析 (appful / Ryusei Okamoto) | iOS | 2026-01-18 | JP store 7,419 ratings, #69–#88 on the Productivity Top Grossing chart. The official site claims over 1M downloads across the series. Estimated monthly revenue about $24K (stringer library). | Category leader. Imports a LINE export txt to produce a 16-type compatibility score card you can share. Later added 擬人化チャット (AI imitates the other person's chat). こいシミュ has no sharing hook, only practice. |
| 恋愛温度計 - もしもの恋愛シミュレーター (Issei Nishi) | iOS | 2026-06-30 | JP store 1,592 ratings, #107–#196 on the Lifestyle Top Grossing chart. Estimated monthly revenue about $4.6K. | **The most direct rival, 1 months earlier.** It also has an AI avatar (「本人みたい」), plus "もしも": rewrite a line from the past and see how the other person would reply. Pricing is nearly identical (weekly ¥600 + credits). Acquisition relied on 2 "@恋愛温度計PR" TikTok persona accounts, each posting 400+ videos. Our research notes include users who said they switched to 恋愛温度計 because こいシミュ cost money. |  |
| Clonely(Issei Nishi) | iOS | 2026-11-03 | Just launched | 恋愛温度計 split the "AI avatar" into its own standalone app, launched in both the Japan and US stores. It competes head-on with こいシミュ's core feature. |  |
| IsTalk - トーク分析&相性診断 | iOS | 2021 | About 7,000 JP ratings | Long-running LINE analysis app. No generative AI reply simulation. |  |
| タラレバDM - もしも〇〇と返信していたら (kotani harumi) | iOS | 2026-08-14 | 0 ratings | You give it a few things the other person often says, and AI mimics their sentence endings, punctuation, sentence length, and message rhythm to reply to you. Nearly the same selling point as こいシミュ, launched two weeks later, didn't take off. |  |
| Reply suggestion apps: スモンド, モテ返信AI, AI返信ナビ, Relora, daptalk, Vibey, モテリプ, Henji, etc. | iOS | 2025-08 to 2026-09 | Mostly 0–3 ratings | All work as "screenshot → AI gives 3 replies", helping you write replies. こいシミュ adds a step: recreate the other person as an AI you can practice with. |  |
| 推しトーク / 推しチャット / Luvlinx / Heyyyy and other AI partner apps | iOS | 2023–2026 | 推しトーク has 920 ratings | They're built around fictional characters, not the real person you're talking to. |  |
| ChatGPT / Gemini, LINE AI Friends | Big-company features | Already there | Huge | Upload the screenshot to ChatGPT and ask it to play the other person, and you can do the same thing. こいシミュ just packages the flow and wraps it in a LINE-style interface. |  |
| RIZZ (US store) and other English rizz apps | iOS / Android | Since 2022 | RIZZ has 68K iOS ratings, estimated monthly revenue about $500K | English-market "AI helps you reply to texts". It offers reply suggestions, not a simulation of the other person. |  |

**Bottom line** : the "screenshot → recreate the other person → practice" idea was not invented by こいシミュ and isn't unique to it. めろとーく's 擬人化チャット, 恋愛温度計's AI avatar and "もしも" all came earlier and are an order of magnitude bigger.

### 3. Growth tactics

| Channel | Evidence | Confidence |
|---|---|---|
| Self-run Instagram Reels account | Official account @koisimyu; the search result shows its original name "恋愛トーク練習 (@koisimyu) - こいシミュ" (Brave search result). Starting from 8-1, it posts Reels tagged #こいシミュ #line #恋愛 #fyp, with LINE mini-skits such as 8-4 "彼氏が夜中に送って、1分で消したメッセージ" (a boyfriend sent a message late at night and unsent it 1 minutes later). The caption is always "▷ 物語に出てくるAIは、プロフィールのリンクから使えます(最初の結果まで無料)" (the AI from the story is available via the profile link; first result free). It also posts tutorial Reels on rewording messages, e.g. 8-1 "受かった!!" rewritten as "ちょっといいことあった". It has also posted MBTI engagement bait, e.g. 8-18 "周りにENFPがいたら、送ってあげて" (if you know an ENFP, send them this) | Confirmed the account is official; view counts not visible (Instagram requires login) |
| Users saw it on Instagram and then downloaded | Review (2★, 9-22): "インスタでいいと書いてあったので入れました" (I saw on Instagram that it was good, so I installed it). The recall says the buzz was "Instagram word of mouth, organic user posts". Our check suggests the content was more likely posted by **its own account** ; we found no third-party creators or organic user posts | Partly confirmed: the source was Instagram; whether it was organic is unconfirmed, though most likely not |
| Stopped pushing later | The same @koisimyu account is now renamed "walkstone - 歩いて磨く", and 11-17 now posts only Reels for Walkstone (10-27, a walking app) | Confirmed |
| The same Reels template also shows up elsewhere | The web product "仲裁くん / HearBoth" (chusaikun.com, operated by CROP., a Japanese SNS company that works on YouTube channels). Its Instagram account 10-15 posts Reels using the identical line "物語に出てくるAIは、プロフィールのリンクから使えます(最初の結果まで無料)" | Identical template confirmed; whether it's the same operator, the same outsourced agency, or copying each other is unknown |
| Store traffic riding on similar category heat | The Japan store's "customers also bought" section includes 恋愛温度計; the store name and subtitle contain "トーク", "AI", "練習" | Inferred |
| Paid ads | Meta ad library (Scrape Creators) returns 0 results searching "こいシミュ", "koisimyu", and "6793537329"; our own metaads library has no record | Nothing found doesn't mean no ads were run |
| TikTok / X / media | x-radar: 0 posts in the past 120 days; TikHub balance insufficient, search failed; Brave and Jina can't find related TikTok content; media coverage is limited to auto-generated app intro pages like APPLION and AppGeo | None found |

### 4. Why it took off (and how far it went)

1. **The demand was created by others** : by 2026, a "feed a LINE chat to an app to see if they like me" habit had already become common among young Japanese. めろとーく built its base by posting score cards on Instagram Stories, with an estimated ~100k downloads in its first month. 恋愛温度計 started in July, using TikTok persona accounts to mass-post LINE chat videos, and made the "AI clone looks like the real person" hook popular. こいシミュ launched 7-29, right after 恋愛温度計 took off (the 恋愛温度計 TikTok account opened 7-04, with a single video reaching 8-03 with 30k views; data from the research notes on 恋愛温度計 in the stringer library).

2. **Its earliest users came from its own Instagram Reels** : LINE mini-skits plus "the AI from the story is available via the profile link", the same playbook as 恋愛温度計 on TikTok, just on Instagram. The only review mentioning a source points to Instagram.

3. **Users stay for the "looks like the real person" effect** : reviews include "再現度高い!" (very faithful, 8-01), "本人すぎて抜けられない" (too much like the real person, can't stop, 10-04), "結構言いそうな事言ってる" (says things they'd plausibly say, 11-02), plus jokes about feeding in an ex's chat history and staging a "修羅場" (showdown) with the current boyfriend (9-19). Negative reviews are almost all about paywalls: "すぐクレジット無くなって話せなくなる" (credits run out fast and you can't talk anymore), "10ラリー…それ以上全然できません" (only 10 exchanges, then nothing), "練習くらいチケットいらんだろ" (practice shouldn't need tickets). One review says it didn't learn the other person's speech habits at all and even confused their gender (10-04).

4. **Hard monetization, just enough to chart on a low-threshold list** : a very small free allowance (about 10 exchanges, with a video ad watched daily to earn more), and it quickly pushes a weekly subscription at ¥600 or credit packs. The threshold for the Japan lifestyle Top Grossing chart is low, so 150 ratings can hold a spot at around #194.

### 5. Differentiation, and is it the reason it's taking off?

- **Three ways it differs from comparable apps** :
    1. **Builds conversations from screenshots** , no LINE txt export needed (めろとーく and 恋愛温度計 mainly rely on exports), so it also works with Instagram DMs and other chat apps, lowering the barrier;
    2. The UI is **rebuilt as a LINE-style chat room** (the 1.4.x release notes say "LINE風トークの再現を見直し", i.e. reworked the LINE-style chat reproduction), where you practice;
    3. **Three-direction polish** before sending (gentle / natural / one step further). Later it added a "恋愛通知表" (relationship report card) that scores the relationship like a school report.

- **Are these why it took off?** Mostly not. The core "AI imitates the other person" idea had several earlier players with volume an order of magnitude larger; screenshot input and three-direction polish were already common in reply-suggestion apps. Its modest traction came mainly from category heat plus self-run Reels traffic, and its "looks like the real person" experience got some users to pay for the weekly subscription. The differentiation is at most "not worse than others", not a real gap.

### 6. Visibility and copycats

- **How visible it is: still barely known.** No media coverage, 0 posts on X, no public revenue, and the founder hasn't shared any numbers; the website is just a landing page on Vercel. The only things found are auto-generated intro pages like APPLION, AppGeo, and AppYonder.

- **Copycats** : hard to say anyone is copying it. タラレバDM (8-14) has nearly the same selling point but didn't take off; Clonely (11-03, the developer of 恋愛温度計) split out the AI clone as its own app and even launched in the US store. That looks more like a leader in the category adding to its lineup than copying こいシミュ.

### 7. What others can borrow

- **Story-style Reels plus "the AI from the story is available via the profile link"** : faceless, mass-producible LINE mini-skit short videos that funnel viewers to an AI tool that reproduces the plot. The template has already shown up in 恋愛温度計 (TikTok), こいシミュ (Instagram), and 仲裁くん (Instagram), which suggests it works for Japanese dating and interpersonal topics. Each product gets a dedicated account, renamed when reused for the next app.

- **Fast follow once the category heats up** : don't go head-to-head on the hottest feature (compatibility scoring); instead pick a side feature (practice partner, pre-send polish) and build it as a separate app, shipping within a month, with pricing copied from the proven weekly ¥600 + per-use tickets.

- **Lessons learned** : the free allowance was too small, negative reviews are almost all about paywalls, and users were poached by similar apps; once the promo account was switched to push a new app, growth stopped. Mass-producing apps means each one never gets deep.

- **Switch markets** : South Korea (KakaoTalk), Taiwan (LINE), and Southeast Asia (LINE / Messenger) have the same habit of asking "does this person like me?" via screenshots. Building "screenshot recreation of the other person + practice + pre-send polish" there, with story-style Reels / TikTok in the local language, hasn't been claimed by anyone as far as I've seen (not verified one by one).

### 8. Still unclear

- Follower count of the Instagram account, Reels view counts, and how many posts it made (not visible without logging in to Instagram); which specific posts drove traffic during こいシミュ's promotion period.

- Where 11-17's single-day +9 ratings came from.

- Whether 仲裁くん and こいシミュ use the same Reels copy because of the same operator and agency, or because they copied each other.

- Whether it was ever promoted on TikTok (TikHub balance insufficient, search failed).

- Real revenue and paid conversion: chart positions only indicate a very small scale.

### Sources

- App Store: [apps.apple.com/jp/app/id6793537329](https://apps.apple.com/jp/app/id6793537329); iTunes lookup (id=6793537329, developer 6791958118, Japan store, 2026-11-18)

- appstore library app_day / app_profile / app_name (10-28 to 11-18, jp)

- Apple Japan review RSS: [itunes.apple.com/jp/rss/customerreviews/id=6793537329/sortby…](https://itunes.apple.com/jp/rss/customerreviews/id=6793537329/sortby=mostrecent/json)

- APPLION (in-app purchase rankings, rating distribution): [applion.jp/こいシミュ/iphone-6793537329/](https://applion.jp/こいシミュ/iphone-6793537329/)

- Website and support: [lovetrick.vercel.app/](https://lovetrick.vercel.app/), [lovetrick.vercel.app/support](https://lovetrick.vercel.app/support); levipon.jp

- Instagram: [www.instagram.com/reel/Dbfm6oDBsgy/](https://www.instagram.com/reel/Dbfm6oDBsgy/) (8-01), [www.instagram.com/reel/DbnWd_7Oqhq/](https://www.instagram.com/reel/DbnWd_7Oqhq/) (8-04), [www.instagram.com/reel/DcLFRJIRPj8/](https://www.instagram.com/reel/DcLFRJIRPj8/) (8-18), [www.instagram.com/koisimyu/](https://www.instagram.com/koisimyu/) (now named walkstone)

- 仲裁くん: [www.instagram.com/reel/Deg9rr6gHBa/](https://www.instagram.com/reel/Deg9rr6gHBa/) (10-15), [chusaikun.com/](https://chusaikun.com/), [crop.tokyo/](https://crop.tokyo/)

- Competitor data: research notes for 恋愛温度計 (ios_6784031888) and めろとーく (ios_6757545135) in the stringer apps library; batch iTunes lookup of Japan-store competitors (11-18)

- Meta ad library (Scrape Creators, 11-18 search for 0 results); x-radar (0 posts)

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