"Pay for the second photo": Couple AI, 150k installs in two and a half months, is a template split out of an AI portrait factory

Couple AI is a Google Play app that merges two selfies into couple portraits. It launched at the end of August, reached 155k installs by Nov 6, and has charted in the Art & Design top grossing lists in Australia, Brazil, Germany and the US. It isn't run by an indie developer. It's a Bermuda company with 9 AI apps and 24M installs in total, and Couple AI is one feature of its flagship product, released separately under its own name. The approach is worth learning; the product itself isn't worth copying.

One bad review, and a reply that missed the point 25 minutes later

On the morning of Sep 28, 2026, a user from South Africa left a one-star review on Couple AI's Google Play page: "Not free, you have to pay once you add a second photo, uninstalled." 25 minutes later the developer replied: "We understand ads can be annoying, but it's the ads that let us keep offering this app for free."

The user was talking about the paywall, but the reply was about ads. The same line later appeared word for word under another one-star review from Oct 27, this one complaining "not free, misleading ads, says 'Try it for free'". Going through their latest 15 reviews, the developer replies to nearly every one, and the replies almost always land 26 or 52 minutes past the hour (Google Play reviews, 9-12 to 10-29). This looks like a scheduled auto-reply bot (inferred, high confidence).

This review more or less spells out how it makes money. The ad says "try it free"; you upload one photo and see the result; if you want a couple photo, you have to upload a second, and that step costs money, $7.99–12.99 per generation (Google Play in-app purchase range, 11-06).

What it is

The first line of the store page reads: "Turn two everyday selfies into studio-quality portraits in seconds." One photo gives a solo portrait, two give a couple photo, in styles including romantic, wedding, engagement, vintage, royal, cinematic and cartoon. Another store page line reads "Dream Wedding Photos — create the wedding shoot you never had."

There's nothing new in the features. ChatGPT's image feature, Gemini and CapCut templates can all do this; the app just packages it into one-tap templates. The version has stayed at 1.3 since Sep 11 and hasn't been updated since.

The numbers: jumping in steps

(Google Play installs and third-party estimates) gained 36k in the last week. The growth curve is uneven: it jumped 31k in a single day in Oct 11 and 18k on Nov 3, and barely moved for several days in between. That looks like paid campaigns run in budget batches, not organic word of mouth (inference).

There are only 72 ratings, less than one per thousand installs. The distribution is split at the extremes: 1 stars for 26 of them, 5 stars for 39, and no 2-star or 4-star ratings at all (11-06). Most people who install it were pulled in by ads, tried it briefly and left, so very few leave a review.

On the charts, its best free-chart rank is #6 in South Africa and #12 in Australia, with Malaysia, Turkey and Poland also in the top 60. Top Grossing runs at #16 in Australia, 12–22 in Brazil, 32–35 in Germany and around #50 in the US (Art & Design category, 10-29 to 11-06). A string of other countries shows Top Grossing ranks of 54–55 every day; such neat numbers shouldn't be read as real revenue for those countries. The four storefronts that actually show people paying are Australia, Brazil, Germany and the US.

Who built it: an AI portrait factory

The developer is Ocean Float Mobile, registered in Bermuda, founded in 2021, with 11–50 staff and no funding record (Tracxn, LinkedIn). Google Play shows 9 apps under it, about 24M installs combined. The main one is MePic (AI photo and video generation) with 8.59M installs, #2 in Australia's Art & Design top grossing chart. Others include the AI music app Beatron (5.26M) and the AI portrait app Lumy (790k).

The key is MePic's one-line description: "AI Photo Editor, Couple & Family Portraits, AI Video Generator". Couple portraits were already a feature of MePic. Couple AI uses the same templates and backend, split out and given its own name, and the package name ai.couple.photo.video.maker is itself a string of search terms.

Its iOS version, "Couple Portrait - AI Photo", only launched in Oct 7 and still has no ratings.

Where users come from

Ads, run by the parent company. In Meta's ad library, MePic's page has 40 video ads running, all on the theme of "family / couple studio portraits", in English, Spanish, Russian and Turkish, and a large batch of new creatives went up from Oct 30 to Nov 3. In TikTok's European ad library, a UK company called OAK SMART TECHNOLOGY (which also advertises Ocean Float's Beatron) posted 33 Android-only ads from Nov 1 to 4, with the copy "Turn blurry snaps into breathtaking couple portraits". From outside, there's no telling whether these ads drive to Couple AI or MePic. Searching Meta for Couple AI's own package name turned up no landing ads. The likely explanation is that the company buys ads by theme, and the couple-theme traffic is split across two packages (inference).

Store search. The title is "Couple AI - AI Photo Generator", and the description is stuffed with couple photo, wedding and engagement. When someone searches "couple ai", an app whose name is exactly that naturally ranks high (inference).

TikTok is creating demand for it, but nobody mentions it. AI couple photos keep coming in waves on TikTok. In late October, a ChatGPT "couple polaroid" prompt video got 129k views; in early November, several CapCut and Hypic "couple back tattoo" templates got 100k to 600k views each; Indian media also wrote up Gemini couple prompts (ET Now, NDTV Profit). These videos promote ChatGPT, Gemini and CapCut. Not one mentions Couple AI. It catches the search and ad traffic next to that demand.

Who it's competing with

(Google Play installs, 11-06) Couple AI is mid-to-low in this group and wasn't first. The category has long been filled by AI portrait factories. The only thing nobody has noticed is this one package.

Why it can work

1. Almost zero cost. Templates, backend, paywall and ad creatives were all proven by MePic. Splitting out a package and launching it gets installs within two weeks, and it doesn't even need version updates afterward.
2. One funnel: a free hook, then charging for the second photo. Nearly all the bad reviews complain about exactly this, yet it charts in the top grossing lists in four countries, so the complaints don't stop the money from coming in.
3. The name is the search term. Inside MePic, couple portraits are just one entry on the home screen. As a standalone Couple AI, ads, store page and paywall all revolve around that one scene, which should convert better.
4. Buy where it's cheap. Its best free-chart ranks are in South Africa, Australia, Malaysia, Turkey and Poland, where category charts are easy to climb and paid acquisition is cheap.

If it were you

Verdict: Don't copy this app itself. It runs on its parent's ready-made product, creatives and ad team, and standing alone it's just a template that ChatGPT could replace at any time. If you build one from scratch, you'll spend as much as it did, without a flagship with 8.59M installs to spread the cost across.

Who it suits: Teams that already run a general AI portrait or photo-editing tool, buy paid traffic, and know which template converts best. What's worth copying for you is the split: take the best-converting single scene, make it a standalone "scene name = app name" package, share the backend and creatives, and gain an extra store-search and ad audience.

Who it doesn't suit: Indie developers without ad experience who want to win on product differentiation, or anyone worried about bad reviews or aiming for a long-term brand. This "free to try, pay for the second photo" play barely scores above 3 and won't build word of mouth.

How to do it: First, find which template has the highest payment rate in your general tool. Split it into its own package and aim the title, package name and description at that search term. Swap ad themes weekly to follow TikTok trends (wedding, polaroid, back tattoo, aging). Start by buying cheap storefronts like South Africa, Malaysia and Turkey to climb category charts, then check the Top Grossing ranks in Australia, Germany and the US.

When to stop: Niche packs like this have a shelf life tied to the trend. CoupleLens went from peak to a few thousand installs a week in under six months, and Emberly has already renamed itself to go broad with general portraits. Cut fast, and move to the next scene once one wave is used up.

What would mean we misread it: If over the next few weeks its Top Grossing rank drops out of the top 60 in the AU, DE and US stores while installs keep climbing, the funnel is just burning money to buy volume without paying back. If the iOS version still has no ratings two or three months from now, the trick depends on the cheap Android traffic.