# No need to open the app: an AI tutor calls you at the scheduled time. Korea's 5-person team Heyring made it into the education Top Grossing chart, and then Speak copied it · Mobile app

> Source: https://stringers.io/en/apps/play_com.hypernovacorp.navatar_android.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

> It turns the "phone English" that Koreans already accept and pay for into a cheaper AI version. Solving the problem of English-learning apps going unused after two weeks, it uses AI that calls you proactively and redials if you don't pick up, then sells annual subscriptions through Instagram ads and creators.

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

헤이링 Heyring: Call & Learn English · HYPERNOVA

# No need to open the app: an AI tutor calls you at the scheduled time. Korea's 5-person team Heyring made it into the education Top Grossing chart, and then Speak copied it

**What it is** An AI phone English app: you set a time, an AI tutor calls you at that time to chat in English, and after you hang up it gives you a report.

**Why it took off** It took "phone English," a format Koreans already accepted and were willing to pay for, and made a cheap AI version. "The AI calls you first, and redials if you don't pick up" solves the problem of English apps going unused after two weeks. Then Instagram ads and influencers sold annual subscriptions.

**B** **How to follow it** Don't build an English version for Korea, Japan or Taiwan, since Speak has already copied that. Instead, take the "calls you at set times + remembers you as an AI companion + charges an annual fee after a 5-minute trial" model to other languages Speak isn't covering, or to daily-habit apps like vocab drilling or fitness, and acquire users with creator videos plus paid ads.

Grade B: "The AI calls you first" is a retention play that has been validated by payments, with almost no technical barrier, so it's easy to replicate. But it has been publicly reported in Korea, Speak and Ringle have shipped the same feature, and it also launched in Japan and Taiwan. Followers can only bet on other languages, markets, or categories, and they need to know how to run paid ads. Publicly known / widely discussed

Data (updated daily) · first launch 2026-04-22 · 11-07 installs 77,395

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.

Heyring (헤이링) is an AI phone English app from HYPERNOVA, a small Korean company of 5 people. You set a time, an AI tutor calls you, and you chat in English for ten minutes; after hanging up you get a report. Through an annual subscription of 98,000 won, it took an app with fewer than 80k Android installs into the Korean education Top Grossing chart. By Oct 20 of 2026, the category leader Speak had launched the same feature. Below, we explain why it took off, how it makes money, how far it has been copied, and where the idea could still be moved.

### Oct 20, two things on the same day

In Oct 20 of 2026, the Korean paper MoneyToday ran an interview with HYPERNOVA founder Park Hyun-mu. He opened with: "Many people feel they should learn English, but about two weeks in, the number of times they open the app falls off a cliff. Learning English feels like homework, and even opening the app becomes a burden. So we flipped it: instead of users opening the app to study, the AI calls them first." He also said something confident: "From a marketing view, this market is very competitive; from a retention view, it isn't."

That same day, Speak (스픽) launched "폰 스픽" (Phone Speak) in Korea, Japan, and Taiwan at once: book a time in the app and an AI tutor calls you; you can also dial it anytime. Top-tier paid users get up to 10 minutes per call, 15 times a month (Chosun Ilbo education edition, news1,10-21). Speak has 12.8M installs on Google Play and ranked 2 on Korea's education Top Grossing chart.

Over the next two weeks, Heyring slid on Google Play's Korea education Top Grossing chart from #16 to #27 (10-29 to 11-06), and its Top Free rank dropped from #91 to beyond #200. The two events lined up in time, but from the outside you can't tell whether one caused the other; it's also possible Heyring cut its own ad spend.

### What it is

You set a time in the app, say 8 every morning. When it hits, your phone rings with a call screen that looks just like a real call. Pick up and you get an AI companion called "Emma". Don't pick up, and it calls again.

What you talk about depends on your life: the AI remembers the interests you mentioned and recent events and picks up where you left off. When you say an awkward sentence in English, it offers the way a native speaker would say it on the spot. After the call you get a report covering fluency, speaking speed, the number of words you used, and key expressions, and you can ask follow-up questions about it. Park says the level of correction should vary by person: people who need to start speaking fast and people who want every sentence perfect want different things.

There's no secret technology underneath: large language models plus speech recognition and speech synthesis. The iOS version launched in Korea in July of 2024 and is the main platform, with 2,407 ratings and a 4.67 rating. At its best it reached #6 on Korea's education Top Grossing chart (App Store Korea storefront, 11-06).

### Why "phone English"

This step was deliberate, not stumbled into. HYPERNOVA was founded in Oct 2022. In its first two years it built and shut down 13 products aimed at individual users, then spent three months interviewing users before settling on phone English (MoneyToday, 10-20).

The logic is simple: phone English is an old category in Korea. For decades, real foreign tutors have called learners at scheduled times, so people already accept the format and are used to paying for it. Meanwhile, mainstream language learning overseas is video lessons. AI versions are much cheaper than human tutors and available 24 hours a day. One Tistory blogger ran the numbers: "At least 30 to 50 percent cheaper than chatting with a real person." He also wrote a candid line: "Lower your expectations a little, and it's a pretty good program."

Users like most that it's passive. The first benefit listed in one Threads review: "Whether I want to or not, it calls me on time, which is good." The second: "I can use it in place of an alarm and start the day with phone English." Another Instagram post said: "The bar for calling a real person for English is too high. With AI, I'm willing to say things even if I get them wrong."

The company says this design produced 14-day retention of 58%, "12 times the average for education apps" (official site, self-reported, unverified).

### Where the money comes from: a 5-minute trial, then an annual-fee prompt

New users can chat for free until about 5 minutes in, when a "special gift" pops up: an annual fee of 98,000 won (about $70). Several Naver blog "bought with my own money" reviews mention this moment. The Instagram profile has long shown a "72% off" banner, and a "77% off" promotion ran over New Year. The store lists in-app purchases between $14.99 and $99.99.

Pushing a one-time annual fee at the peak of the experience raises the average order value and speeds payback. The founder has publicly shared a string of numbers: about 4 months after launch, "two-week revenue of 10M won, ROI 560%" (LinkedIn, 2025-01); "revenue passed 100M won in 3 months, with 20% of monthly active users paying" (DIO blog, 2025-04); and the hiring page says "No. 4 in Korean education app revenue, operating margin 52.6%." All of these come from the company itself and can't be verified. Funding has been a single seed round in July of 2023, and it has run on its own profits since; Park says they'll consider raising once annual revenue reaches 10B won.

Ranking #27 on Google Play's Korea education Top Grossing chart with fewer than 80k Android installs shows that few people pay, but each one pays a lot.

### Where users come from

| Channel | What we saw |
|---|---|
| Instagram / Meta paid ads | The founder said "paid marketing is the most effective way to validate selling points for education apps." An outsourced marketing team produced ad storyboards and video creatives. ROI of 560% refers to the return on that ad spend. |
| Instagram influencers | Many seeding Reels tagged #광고 (ad), plus partnership posts giving away monthly passes in giveaways. |
| Naver blogs, Tistory, Threads | A batch of "내돈내산" (bought with my own money) reviews; one English teacher said on Threads that it is "very popular in English conversation teacher circles." |
| Store search | The iOS title is "헤이링: AI 전화영어, 스픽·영어회화·스피킹·발음" (Heyring: AI phone English, Speak·English conversation·speaking·pronunciation), which writes the rival Speak's Korean name "스픽" directly into its own title. |
| Press | Daily Economy (2025-07-30), MoneyToday (2026-10-20) |

How much was spent on ads can't be found. Almost nobody discusses it in English or Chinese online, and it can't be found on TikTok either.

### Who it's competing with

Its only differentiator is "AI that calls on time, plus a friend who remembers you." That is being copied one piece at a time:

| Competitors | When | Scale | How it differs from Heyring |
|---|---|---|---|
| Speak "폰 스픽" | 2026-10-20, launched in Korea, Japan, and Taiwan at once | 12.8M Play installs, #2 on Korea's education Top Grossing chart | The category leader with a complete course system. Scheduled calls, call-anytime, and memory, nearly copied one-to-one. |
| Ringle AI 전화영어 | 2025-08 | Korean human tutors, top of ${1:1} English | AI calls first at the agreed time, 4 accents, uses in-house materials |
| Weverse Brain 맥스AI 전화영어 | 2026-09 | 300 corporate clients | Sold to companies for staff training, 10 minutes per call, redials if no answer |
| Praktika, Loora, Learna AI, YayTalk | 2022–2026 | Hundreds of thousands to tens of millions of installs | AI speaking practice partners that users must open themselves; no calls |
| 영어하자 - AI 전화영어 | 2026-03 | iOS #5 ratings | Small copycats from individual developers |
| EnglishCall AI and others | 2024–2026 | Small | Calls you on a real phone number, focused on the English market, from $24.99/month |
| ChatGPT, Gemini voice mode | From ¥2023 | huge | Free and general-purpose, but doesn't call you first, doesn't push you, and has no report |

In Korea this spot is already crowded: the leader, the top human tutors, and corporate training have all built the same thing. Heyring's remaining cards are the relationship and memory built with its AI companion, a lower price, and an early move overseas. But going overseas isn't easy either: it has launched in Japan and operates in Taiwan, yet on iOS it has only 24 ratings in Taiwan and 3 in Japan, while Speak's phone feature launched in both places on day one.

One thing Park learned in Taiwan is worth noting: selling it as "like a call with a friend" converts better than selling it as "phone English." The same product needs a different selling point in a different market.

### If it were you

**Verdict:** Worth trying, but don't follow into Korea, Japan, or Taiwan. "AI calls you at the scheduled time" is a validated retention play with almost no technical barrier. That's exactly the problem: Speak copied it in one release. What you can do is go to the markets and audiences that Speak and Ringle haven't gotten to.

**Best suited for:** A team of one or two people that can build a smooth voice-call app, has experience running paid ads in a specific native-language market (or a specific audience), and can write ad copy and influencer scripts that sound like a real local.

**Who it's not for:** Anyone trying to do English speaking practice in Korea, Japan or Taiwan (Speak already has a same-style feature, and so do Ringle and a crowd of AI tutors); anyone who can code but can't run ads (Heyring says paid acquisition is its main way of validating the pitch); anyone counting on a tech edge (the underlying stack is open to everyone).

**How to do it:** Three paths, from most to least certain: ① Switch languages and build the "learn other languages" version. People learning Korean, Japanese or Spanish have the same "I stop opening it after two weeks" problem, and per October's report, Speak's phone feature only shipped in its English courses in Korea, Japan and Taiwan; ② Switch markets, to places where Speak is weak and there's a tradition of "foreign teacher phone calls," like Southeast Asia and Latin America, which have Filipino teacher phone lessons and online lesson habits (inference, you'd need to check local charts for the same product first); ③ Take the "incoming call" mechanic out of language learning: vocab drills, fitness check-ins, meditation, companionship calls for seniors, anything that needs daily consistency and that users would never open on their own. On product, copy three things: a realistic incoming-call screen with redial if you miss it, an AI companion that remembers you, and a one-page report after hanging up. For pricing, copy the one-time annual discount that pops up after the free trial reaches 5 minutes.

**How to get users:** Creator trial videos on Instagram and TikTok, filmed around the few seconds of "alarm goes off, you pick up, it's your AI tutor." Pair them with local paid review blogs. A common move is putting the category leader's name in the store title, like it does (mind the platform rules). Test with a small budget first and see whether trial-to-annual conversion lets paid acquisition pay back, then spend more.

**When is the window:** Speak has already made the phone feature standard, and its next step is likely adding more languages and markets. You'd need to get paid acquisition to break even in one market within three to six months, or big players will close the window too.

**What would prove you're wrong:** ① Heyring climbs back into the top 15 on Korea's Top Grossing chart and its iOS rating growth recovers, which would mean relationships and memory are a real moat, and Speak copying the feature can't copy the users, so following it in Korea is still not too late; ② Speak rolls the phone feature out to Korean, Japanese, Spanish and other courses and more countries within a few months, which closes paths ① and ② above; ③ Someone launches the same product in Southeast Asia or Latin America, spends a month or two on it, and still can't crack the local education Top Grossing chart, which would mean the approach depends on Korea's habit of paying for "phone English," an old category, and can't be moved.

### Data and sources

- Google Play: installs (10-29 to 11-06,11-06 is 77,395, last 7 days +1,098), Korea education Top Grossing and Top Free ranks, in-app purchase range, store description; installs and ranks of same-chart competitors (11-06).

- App Store Korea storefront: launch date, rating count and average, best rank; rating counts in the Taiwan and Japan stores (11-06).

- MoneyToday interview (2026-10-20), Daily Economy (2025-07-30); DIO blog (2025-04); founder's LinkedIn; official site and careers page. Revenue, retention and margins are self-reported by the company and unverified.

- Speak "폰 스픽": Chosun Ilbo education section, news1 (2026-10-21), Speak's official Instagram.

- User reviews: public reviews and sponsored posts on Tistory, Naver blogs, Threads and Instagram.

- Where it says "inferred" or "unverified," it means inference or unchecked.

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

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

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| Speak 스픽 (including 폰 스픽) | iOS | 2019; phone feature 2026-10-20 | 12.8M Play installs, #2 on Korea's education Top Grossing chart | Category leader with a complete course system; the phone call is just one feature. You can schedule calls or call anytime, it also has memory, and it directly copies Heyring's core experience |
| Ringle AI 전화영어 | Android | 2025-08 | Korean live-action 1:1 English brand leaders | The AI calls you first at a set time, offers 4 accent options and uses its own textbooks; it's an add-on to the main product |
| Weverse Brain 맥스AI 전화영어 | Other | 2026-09 | 300 corporate clients | Built for corporate employee training, 10 minutes per session, not a consumer subscription for individuals |
| Praktika / Loora / Learna AI / YayTalk | Android | 2022–2026 | 2M to 54M installs | AI speaking practice that requires the user to open the app; no proactive calls and no friend-like relationship |
| Live phone English (Minbyeongcheol, Catchit and others) | Other | Already existed | Mature category in Korea | Live teachers, expensive, fixed times; Heyring is the AI version, callable anytime for 24 hours, at a fraction of the price |
| ChatGPT / Gemini voice mode | Big-company features | From ¥2023 | huge | Free and general-purpose, but it won't call you proactively, has no learning report, and doesn't nudge the user |
| EnglishCall AI and others | Web | 2024–2026 | Small | Uses real phone numbers, aimed at English-speaking markets, from $24.99/month |

Where users come from

- **Confirmed** **Instagram/Meta paid ads** : the founder says "paid marketing is the most effective way to validate the pitch." An outsourced marketing agency set up ad dashboards and produced video creative (DIO blog 2025-04); LinkedIn shared an ROI of 560%

- **Confirmed** **Instagram creator Reels** : several Reels tagged #광고 (ad), plus a giveaway post offering monthly passes

- **Confirmed** **Naver blog and Threads word of mouth** : several "내돈내산" (bought with my own money) reviews; on Threads an English teacher says it's "hugely popular in English conversation teacher communities"

- **Confirmed** **Store search ASO** : titles packed with keywords like "AI 전화영어, 영어회화, 스피킹, 발음" (AI phone English, English conversation, speaking, pronunciation); the iOS title also includes the competitor's name, "스픽"

- **Confirmed** **Media and startup circles** : Daily Economy 2025-07-30 and MoneyToday 2026-10-20 interviews; the founder posts revenue on LinkedIn

- **Confirmed** **Going global (Japan, Taiwan, Southeast Asia)** : 24 ratings on the iOS Taiwan store and 3 in Japan; on Android it charted in the education Top Grossing lists for Hong Kong, Malaysia, Vietnam and Thailand, ranking beyond #10-29 to #190

## Why Heyring (AI phone English) managed to break out

**In one line** : an AI phone English app built by a Korean team of 5 at HYPERNOVA. The AI tutor calls you at the times you set and chats with you in English; after the call you get a fluency and rephrasing report. It makes money from annual subscriptions (a common Korean discount price is 98,000 won/year). **Verdict** : It taps into "phone English," a learning format Korean users already know, and swaps "user opens the app" for "AI calls at a set time," which fixes the retention problem that kills most education apps. Add Instagram ads and creator promotion to sell annual subscriptions, and a small team got into the top of Korea's education Top Grossing chart. From 2026 to October 9, Ringle, Weverse Brain and category leader Speak each shipped the same feature, and that differentiation is being erased.

### 1. Data (from our database)

- **Google Play** (googleplay database, 2026-11-06): installs 77,395, last 7 days +1,098; the data card has no value for the 30-day change. Most revenue is in Korea. The launch date in the app_profile is 2026-04-22, while the data card shows 2026-09-10, likely reset when the listing info was changed or a storefront was added. Its Korea education Top Grossing rank slid steadily: 10-29 to 10-31, then 16,11-01, 19,11-02, 24,11-06, 27. Over the same period it fell from 91 to 198 on the Korea education Top Free chart before dropping off. On 10-29 it was still on the education Top Grossing charts in Hong Kong, Malaysia, Vietnam and Thailand, ranked 190–195 (chart_rank).

- **iOS** (appstore database): id 6511246156,2024-07-30 is listed in Korea and is the main platform. Korea rating count 2,407, average 4.67, best rank #6 on the education Top Grossing chart. In the last 7 days it added only 4 ratings (10-28 to 11-06 at 0–2 per day), so new downloads have slowed. It has 24 ratings in Taiwan and 3 in Japan, so going global is still very small.

- Neighbors on the same chart (googleplay, kr education Top Grossing, 11-06): #1 Duolingo, #2 Speak (12.8M installs), #3 말해보카, #5 Praktika, #6 Loora, #8 YayTalk, #11 Learna AI. Heyring ranks #27 with under 80k installs. High-priced annual subscriptions validate its depth of monetization.

- Same developer name, different company: the "Hypernova Limited" in the database makes wallpaper and battery animation apps (Silly Fun Smile and others), and it is not the same entity as Korea's "HYPERNOVA / HYPERNOVA Corp."

- Ad library: no Heyring Page found in metaads. That doesn't mean it hasn't advertised, only that it isn't on our tracking list.

### 2. Why it took off

1. **形式选对了:韩国本来就有「电话英语」市场** 。创始人朴炫武(Park Hyun-mu)说,头两年先后做了又关掉 13 个 B2C 产品,又做了 3 个月用户访谈,才确定做电话英语。他的理由是:电话英语在韩国是成熟的学习方式,海外主流则是视频课(MoneyToday 2026-10-20 专访,[www.mt.co.kr/en/future/2026/10/20/2026091814012981961](https://www.mt.co.kr/en/future/2026/10/20/2026091814012981961) )。AI 版的价格只是真人电话英语的零头,还能 24 小时随时打。

2. **Using "incoming calls" to fix retention** . The official site says users don't have to open the app themselves; the phone calls them, and it keeps redialing if they miss it. It claims Day 14 retention of 58%, "12 times the education-app average" (https://www.hypernovacorp.com/, self-reported, unverified). Blog users also say "the most freedom with time" and "AI doesn't make you embarrassed" (https://blog.naver.com/PostView.naver?blogId=vlossum&logNo=224093483250, [shahan26.tistory.com/644](https://shahan26.tistory.com/644)).

3. **The AI companion remembers you and feels like a friend** . The AI character (e.g. "Emma") remembers your interests and recent events. After each call it gives a report (fluency, speaking speed, word count, key expressions) and rewrites what you said the way a native speaker would, and you can ask follow-up questions about the report (same MoneyToday article). The founder says in Taiwan, "talking like calling a friend" converts better than "phone English."

4. **Pricing design: annual offer after 5 free trial minutes** . When the user reaches 5 minutes, a "special gift" pop-up offers an annual plan at 98,000 won (https://m.blog.naver.com/the_maya/224134318278, [m.blog.naver.com/bomnal34/224238813832](https://m.blog.naver.com/bomnal34/224238813832)). The Instagram profile has long displayed a "72% off" banner, and a "77% off" promo ran over New Year (founder's LinkedIn). The in-app purchase range listed in the store is $14.99–99.99.

5. **Self-reported results** (all from the company or founder, unverified): 2024-08 MVP launched, and about 4 months later, in 2025 January, "two-week revenue of 10M won, mostly organic, ROI 560%" (founder's LinkedIn); "revenue passed 100M won within 3 months of launch, 20% of MAU are paying users" (DIO blog 2025-04, [blog.dio.so/company-review-hypernova/](https://blog.dio.so/company-review-hypernova/)); "No. 4 in Korea education app revenue, up 3× in the last 3 months, operating margin 52.6%" (official site, Wanted and GroupBy careers pages). It has raised money once: a 2023-07 seed round from SpringCamp, after which it joined TIPS and Google for Startups, and it has run on its own profits since.

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

| Channel | Evidence | Confirmed? |
|---|---|---|
| Paid social ads (mainly Instagram/Meta) | In a DIO blog post the founder says "paid marketing is the most effective way to validate the pitch for education apps." An outsourced marketing agency built ad dashboards, video creative and creator research; the New Year LinkedIn post mentions social ads; the ROI of 560% refers to the return on that spend | Confirmed (actual spend unknown) |
| Instagram creators and promo Reels | Several Reels tagged "#광고" (ad), such as instagram.com/reel/DWqstKUDoPZ, DRRmspmkTx4; plus a giveaway post offering monthly passes, DRejAbPgd2K | Confirmed |
| Naver blogs, Tistory, Threads word of mouth | Several "내돈내산" (bought with my own money) reviews; on Threads an English teacher says it's "hugely popular in English conversation teacher communities" (threads.com/@english_dori_) | Confirmed (some may be paid creator partnerships) |
| Media and startup circles | Daily Economy 2025-07-30 report, MoneyToday 2026-10-20 interview, founder's LinkedIn revenue posts; a repost on X (@ministerbtc, 2026-10-25,212 views) | Confirmed, but small in scale |
| Store search | Store titles packed with keywords: "AI 전화영어, 영어회화, 스피킹, 발음" (AI phone English, English conversation, speaking, pronunciation); the iOS title also includes "스픽" (Speak), riding on the competitor's name | Confirmed (using Speak's name in the title) |
| Going abroad | Live in Japan and Taiwan, preparing the US and South America; on Android it has charted near the bottom of education Top Grossing lists in Hong Kong, Malaysia, Vietnam and Thailand | Confirmed, very small scale |

### 4. Competitors

| Name | Where | When | Scale | How it differs from Heyring |
|---|---|---|---|---|
| Speak 스픽 (incl. 2026-10-20 newly added 「폰 스픽 / スピーク電話 / AI家教通話」) | iOS/Android | 2019; phone feature 2026-10-20 launched in Korea, Japan and Taiwan at the same time | 12.8M Play installs, #2 on Korea's education Top Grossing chart | Category leader with a full course system. The phone call is just one feature: you can book incoming calls, call anytime, and it remembers you (covered by news1, chosun, prtimes) |
| Ringle AI 전화영어 | iOS/Android | 2025-08 | Ringle is a Korean live-tutor English app, 1:1 at the top | The AI calls at the agreed time. Pick from 4 accents, using Ringle's course materials. Almost identical to Heyring, just an add-on feature on top of its main product |
| Weverse Brain 맥스AI 전화영어 | B2B | 2026-09 | 300 corporate clients, 1.23M members | Sold to companies for employee training. 10 minutes per call, redials if you miss it |
| Praktika / Loora / Learna AI / YayTalk / 톡미 TalkMe | iOS/Android | 2022–2026 | Hundreds of thousands to tens of millions of installs | AI speaking practice you open in the app yourself. It doesn't call you |
| Live phone English (Minbyeongcheol, Catchit and others) | Offline / App | Already around | Mature market | Human teachers. Expensive, and fixed schedules |
| 영어하자 - AI 전화영어 | iOS | 2026-03 | 5 ratings | A small indie knockoff |
| EnglishCall AI and other overseas apps that call you on a real phone number | Web | 2024–2026 | Small | Runs on real phone lines, targets the English market, from $24.99/month |
| ChatGPT / Gemini voice mode | Big-company features | From ¥2023 | huge | Free and general-purpose, but it won't call you, has no learning reports, and won't nudge you |

### 5. The differentiator: is this why it took off?

The only real differentiator is "AI calls you first + a friend-like companion with memory." There's no technical moat: it's all general-purpose LLM + TTS/STT under the hood. Its value is that, in the "phone English" category Koreans already accept, it fixed the problem that made users quit, lifting retention and annual-plan conversion. So this differentiator really is the main reason it took off. Paid acquisition works too, because retention is high enough for the ROI to make sense. But the differentiator is already being copied: Ringle (2025-08), Weverse Brain (2026-09), and Speak (2026-10-20) all shipped the same feature. After Speak launched its phone feature, Heyring fell from 16 to 27 on the Android Korea Education Top Grossing chart, dropped out of the top 200 on the Top Free chart, and its iOS rating growth nearly stalled. The two events line up in time, but the causation is unverified. Going forward its moat rests on: the relationships and memory built up by its AI companions, a lower price, and getting to Japan, Taiwan and other markets before Speak does.

### 6. How visible it is

Fairly public in Korea: the founder posted revenue and ROI on LinkedIn, the careers page says "education app revenue #4, profit margin 52.6%", and Daily Economy and MoneyToday have both covered it, and the category leader Speak has already followed suit. In English and Chinese circles almost nobody talks about it: x-radar's local corpus has no related posts in 90 days, and TikTok search turns up nothing.

### 7. Takeaways

- The "anti-open-rate" design: calls, a real phone UI, and redials turn learning from something users must actively do into something they just answer. It works for any category that needs daily habit (fitness check-ins, vocab apps, meditation, minor languages).

- Pick a offline or traditional learning format the local market already accepts (phone English in Korea), then use AI to make it cheap. That's easier to get people to pay for than inventing a new format from scratch.

- Once the trial hits 5 minutes, pop an annual-plan discount at the peak of the experience. Stuff keywords into the store title and ride competitor names. Instagram creators seed it with #광고 tags.

- Entering other markets means changing the pitch: in Taiwan, "like a friend calling you" works better than "phone English."

### 8. Uncertain

- Revenue, retention, and "revenue #4" are all self-reported by the company, with no independent verification.

- The causal link between Speak's phone launch and Heyring's chart drop is unverified. It could also be seasonal, or the company may simply have cut its ad spend.

- Ad spend and creative volume were not found (metaads doesn't track it).

- Actual revenue from Japan, Taiwan and Southeast Asia is tiny, so it's unclear whether going global can take off.

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