Cal AI

An iPhone app for makeup. The user uploads a selfie without makeup and a photo of a makeup look, for example from TikTok or Instagram. The app's AI applies that look to the user's own face, then gives a step-by-step routine and a list of products to buy. It is free to download and paid for with subscriptions through the App Store.

Its result is a before-and-after picture of the user's own face, which a short video shows in seconds. The founder chose TikTok as the channel before he built the app, then tripled revenue from almost the same installs by moving and testing the paywall with Adapty.

An iPhone and Android app that estimates calories, macronutrients and micronutrients from a photo of a meal, with a barcode scanner for packaged food and a log of weight and BMI. It is free to download; the service needs a subscription after a 3-day free trial, at US$9.99 a month or US$49.99 a year in the US store.

One founder released it in August 2025, after Cal AI, released in April 2024, had shown that photo calorie counting sells. It has a hard paywall, a short trial and cheap yearly plans. It appears to have started in France: the French App Store shows 598 ratings against 138 in the US store. The owner lists TikTok as the only marketing channel.

An iPhone app that coaches boxing. The user uploads a video of shadowboxing, bag work or sparring, and the AI scores the technique in seven areas (stance and balance, guard, footwork, punch selection, combinations, head movement, range control) with corrections and drills. It also has an AI chat coach, training programs, a punch callout mode with audio cues, a round timer, AI food logging from a photo, a barcode or voice, and a weight-cut mode. Every feature needs a subscription.

It moved the structure of the founder's first app, Wrestle AI, to boxing, a sport that people can train alone: an AI breakdown of a training video, a calorie tracker and a weight-cut mode, all behind a subscription at the same prices. Users came from Meta ads and influencers, according to the founder's TrustMRR profile, with a self-reported marketing spend of US$10,000 a month.

A website and phone app that masters a song from a written request such as "warm, powerful and wide stereo". Each request comes back as a new revision that the user can compare with the earlier ones, and a Spotify link can serve as the reference for loudness and tone. The same subscription includes a splitter that separates a song into vocals, drums, bass and instruments. Since September 2026 the company also offers a generator that makes a music video from the song. Unlimited use costs US$16 a month or US$48 a year on the web.

For its first 20 months the product brought in about US$3,500 a month. Then three things came together: phone apps with weekly plans from August 2025, web plans with unlimited use at US$4 a month billed yearly, and a new group of customers, people who make songs with AI generators such as Suno and need the last step done for them. Monthly revenue rose 7.2 times between January and July 2026. No source states which of the changes caused the rise.

A phone app that scores a selfie. The user uploads a photo of his face, and the app returns a score up to 100 with ratings of the jawline, the cheekbones, the skin and masculinity, and tips such as starting a skin care routine. The results are shown after the user subscribes, at about US$4 a week, or invites three friends.

Young men were already posting photos online and asking strangers to rate their looks. Umax turned that habit into a private score from an AI model. One founder built it at almost no cost on a rented vision model, and creators in one small niche spread it, because a score makes a before-and-after video easy to show. Revenue was highest within months of the launch and has fallen since.

A church connects the audio of its service, from a microphone or the sound desk, to Glossa in a browser and starts a stream. Glossa turns the speech into text, translates the text, and reads the translation aloud with a synthetic voice. Each visitor scans a QR code, picks one of more than 100 languages on a phone, and listens with earphones or reads the captions. The church needs no receivers and no interpreter booth. It pays for the minutes of speech in each language, not for the number of listeners.

Churches with members from many countries meet every week, and the older solutions cost too much for most of them: an interpreter for each language, or radio receivers to hand out and collect. Glossa charges US$5 for one hour of speech in one language, counts by the minute, and charges nothing for silence, songs or extra listeners, so a small church can start with four free hours and then a few dollars a week. The product is shaped for this one buyer: captions go into ProPresenter and OBS, which churches already run, and a church can fix how names, Bible quotations and its own terms are translated. A very small team can offer this because the speech recognition and the voices are rented from model companies and paid by use.

An app for iPhone, Android, Mac, Windows and the web that records meetings, phone calls and lectures, writes a transcript with the speakers identified, and then a summary with action items. A user can also ask questions about a recording. It is free for 30 minutes a month; Pro costs US$100 a year in the app stores.

It reached the App Store early, in July 2023, with the simplest form of the idea: press one button, get a transcript and a summary. The founder paid for Meta ads while they still worked for a new kind of app, tested paywalls constantly and moved buyers to annual plans. He works alone with AI coding tools, so in his own 2026 account about US$3 million of about US$7 million a year stays as profit.

An app for iPhone, Android, Mac, Windows and the web that records speech and returns finished text: a clean note, an email, a list, a social post or a meeting summary split by speaker. A phone keyboard and a desktop dictation tool let users speak into any other app. It handles more than 90 languages and recordings of up to 90 minutes.

The first apps that combined OpenAI's Whisper and ChatGPT for voice notes appeared in 2023, and the founder found them "not implemented well enough". His team built a simpler one, collected early users and feature requests through lifetime deals on AppSumo in 2024, and in 2025 grew with advertising and paywall tests measured in Adapty's subscription data.

An iPhone app that turns a voice memo of 30 seconds to 2 minutes into an Instagram or LinkedIn carousel, a reel or TikTok script, a thread for X, story scripts and hooks. It places the text on templates and exports PNG images or a PDF. It can also learn the user's writing style, suggest post ideas, and turn a pasted video link into the user's own carousel.

A carousel needs both writing and design. Draft AI asks for a minute of speech instead and returns finished slides, plus scripts, threads and hooks from the same recording. It charges before the first result, so everyone who uses it has subscribed or started the trial of a paid plan.

You describe an app in plain language. Base44 generates a working web app with its own database, user login and hosting, so you never touch a server.

It launched in the month that AI models became good enough to write a whole app. It sold to people who cannot program, it charged from day one, and it was sold to a company that already had millions of customers.

A web app for real estate agents and photographers. The agent uploads a photo of a room, chooses the room type and a furniture style, and gets back the same room with furniture in about 10 to 15 seconds. It can also remove existing furniture. Plans are monthly subscriptions sized by the number of photos, and the top plan includes an API.

In 2023, general image generators that were asked to furnish a room also changed the room: the founder said DALL-E might turn a window into a wall painting or add a door, which misrepresents a property. VSAI's own model finds the floor, the ceiling, the mirrors and the doors first and only then adds furniture. It cost US$12 to US$69 a month in early 2024, while VSAI's comparison page lists other staging services at US$16 to US$29 per photo with 12 to 48 hours of waiting. Zillow, the listing portal where such photos end up, bought it.

You speak into your phone or browser. Talknotes transcribes the recording with a speech-to-text model and rewrites it in a style you pick: a clean transcript, a short note, a task list, a formal email, a blog post or a video script. It works in more than 50 languages and exports to text, email or a link.

A marketer who had learned to code one year earlier built it in a week, found that paid ads could buy customers for about US$50 each with a yearly plan that paid the ad bill up front, and sold the business for cash eleven months after launch, when the bugs and the support load became more than one person wanted to carry.