A 3D mannequin boxer, 149M views: BoxMaxxing turned the coach in its app into a TikTok account

BoxMaxxing is an AI boxing home training app from a small Australian studio, and you can't train without subscribing. It hasn't been covered by media, nobody mentions it on X, and AI boxing app reviews don't write about it. But its TikTok account grew to 1.37M followers in three months, and Android installs rose with it from a few hundred a day to several thousand a day. Its playbook is simple: the 3D coach in the app is also the star of every video on this account. The same playbook could be run again with another sport, and in boxing someone is already copying it.

A 17-second video

On Sep 20, the TikTok account @boxmaxxing.app posted a 17-second video with only one line of caption: "Try this boxing combo." The frame is black, with a grey metallic-looking mannequin boxer in red shorts and red gloves throwing a combo. No real person, no voiceover, and no mention of the app.

By Nov 6 the video had 19.1M views and 3.31M likes. The top comment is in Spanish, "yo en mi cuarto" (me in my room), 97k likes, with a face gesturing in a bedroom. Third place is "upper+jab+hook+upper+cross+hook ok let's try," 44k likes. Scrolling down, there's Portuguese "Eu no meu quarto," a Russian asking "what's the difference between a cross and a jab?", Vietnamese and Indonesian comments, and a Spanish speaker saying "I'm probably the only girl interested in this." Some also spotted that this is a selling account, writing "I'm not going to pay for the app." 1,568 likes.

That same week, BoxMaxxing's daily Google Play installs went from about 400 in mid-September to between 2,300,9-20 and 9-25 per day around Sep 19, with 1,600 to 5,600 per day (third-party estimates plus Play's public install counts).

What it is

BoxMaxxing is developed by MetaTensor. The company was registered in South Australia in 2024 May, its website calls itself an "independent studio," and it has five apps, four of which are barely used. The BoxMaxxing iOS version launched in Jul 16, and the Android version in Jul 23.

When you open the app, it asks a few questions, then builds a 30-day boxing camp for you. Each day is one lesson: a 3D boxer demonstrates round by round on screen, and new techniques can switch to a slow-motion Teach mode. The front camera counts your punches and scores your moves; there are weekly leaderboards for city, friends, and global. The Nov 4 2.6.0 version added replay recording and supports five languages: English, Spanish, Japanese, Chinese, and Arabic.

It has a hard paywall: you must subscribe to Pro to start training, with weekly, six-month, and annual tiers, priced on Play at $6.99 to $49.99.

Where the volume comes from

Almost entirely from this TikTok account. The account posted its first video on Aug 17, "How to Throw The Perfect Jab," with the bio "Boxing app here 👇" and a link to boxmaxxing.com/tt, a landing page built specifically to catch TikTok traffic. By then the app had been live for a month, and almost nobody on Android had installed it.

Every step on the install curve lines up with a hit video. None of the videos carry a paid-promotion label. Whether it bought user acquisition elsewhere can't be seen from outside, but the inflection points follow the videos, so organic traffic was the main driver. Roughly, every 1,000 views brought in about 0.6 Android installs (estimate); iOS installs aren't counted.

Posting cadence also followed the data: 11 posts in the second half of August, 21 in the second half of September, 26 in the first half of October, and at peak up to 2 to 4 posts in a single day.

What the videos look like. Every one features the same 3D mannequin boxer, on a black background or a neon grid, with captions like "Level 1 / Level 2". There are three formats: graded tutorials on combos, footwork and slipping; challenges like "Try this combo"; and recreations of famous fights, like Mighty Mouse's flurry, McGregor's knockout of Aldo, and Tank Davis's KO, all with this mannequin.

These videos can be mass-produced because the boxer in the video is the coach in the app. The store page reads "Follow a 3D boxer demonstrating every move". The moves, model and camera work already exist in the product, so you can cut a 15 to 20 second clip and post it, with no real person on camera and no shoot. Viewers who watch want to throw the punches too, and the app sells exactly that: "train with this guy".

The traffic is global, not American. The comments are mostly in Spanish, Portuguese, Russian, Indonesian and Vietnamese. Android installs are close to 90k, but the iOS US store has only 151 ratings. On Google Play it appears only in the fitness free charts of the Philippines, Indonesia and South Africa (ranked 150 to 200). On iOS it made the fitness free charts in 48 countries and the fitness grossing charts in 13, peaking at #30, in markets like Azerbaijan, Uzbekistan, the UAE, Brazil and Peru. It charted in none of the US, UK, Canada, Australia or Germany (11-06). Most rival AI boxing apps are iOS-only and English-only. BoxMaxxing is on both platforms in five languages, so the global Android users TikTok sends can actually install it.

Other channels play a small role. The name borrows the Gen Z "-maxxing" slang, which gives it some meme value. In the US iTunes store, it ranks #2 for "boxing training" and #5 for "boxing workout", so store search brings in some volume. It has no SEO, press, Reddit or X presence.

Where the money comes from

The money comes from subscriptions, but the amounts aren't visible from outside. It charted in the iOS fitness grossing charts of 13 countries, which means people are paying. But these are small markets with low chart thresholds, so the chart positions don't reveal revenue.

There's one risk. The latest US review (11-02, one star) reads: "Paid $40, and the camera barely picks up my punches when I'm sparring... don't buy." A hard paywall plus inaccurate detection will eventually produce refunds and bad reviews. Of the 892 ratings on Play, 2-star and 3-star ratings are both 0, which may mean the app only prompts happy users to rate (inference, unverified). So the 4.6 rating shouldn't be taken at face value.

Who it's competing with

AI boxing coaches have been a fairly hot niche for 2026 years. One industry review counted 2025-08 to 2026-05 new ones, 5 in total: 4 were iOS-only and 4 were built by solo developers.

On features alone, BoxMaxxing isn't better than BRAWLER, Boxer AI or PunchLab. What sets it apart is distribution: it's the only one that turned its 3D coach into a million-follower content account, and it launched on Android at the same time.

Someone is already copying it. GlovioAI (@glovioaiapp) hasn't launched its app yet and only has a waitlist. Since October it has posted nearly identical 3D mannequin sparring videos, growing to 75k followers in a month, with one 10-26 and one 10-27 video at 640k and 520k views. Another app, Boxing AI: Ring Coach, renamed itself JabMaxx at the end of September and switched to AI animated videos in October, seemingly imitating it (inference).

If it were you

The verdict: worth following, but don't build another boxing app. What's worth taking is the funnel: "the 3D coach in the product = short-video material", applied to a sport nobody has done this way yet. Boxing's format is already being copied, and a third entrant will struggle to get the traffic September did.

Who it suits: a small team of 1 to 3 people, with someone who can make 3D characters and animation (Blender / Unity, and able to use mocap data or AI motion tools), and someone who can cut 2 to 4 short videos a day. Ideally you've trained in the sport you pick, so you know how to split it into "Level 1 to Level 5".

Who it doesn't suit: anyone counting on AI motion detection to stand out. BoxMaxxing's own bad reviews are about inaccurate detection, so features aren't what decides this business. It also doesn't suit anyone without 3D skills who plans to hire real people to film tutorials, since that caps output at a few videos a day and rules out fight recreations.

How to build it: pick a sport whose moves look good, split into levels and trainable at home: Muay Thai, solo Brazilian jiu-jitsu drills, taekwondo kicks, self-defense, basic street dance steps, kettlebells. Start with a 3D coach and 30 lessons, since the lessons themselves are your footage library. Launch on both platforms and offer Spanish, Portuguese and Indonesian from day one, since that's where the traffic will come from. Use a hard paywall with a weekly subscription, which also collects money in cheap-traffic countries. Camera counting can start as "good enough", and the app shouldn't promise detection accuracy behind the paywall that it can't deliver.

How to get users: run a faceless main account posting only 3D coaching, 15 to 20 seconds per clip, rotating three formats: graded tutorials (Level 1→5), challenges ("Try this combo"), and recreations of the sport's famous fights with the 3D model. Use a one-line signature, link the profile to a dedicated landing page so you can track conversion per channel, and repost the same videos to Instagram Reels. Don't push for installs in the first month. BoxMaxxing also posted for a month before a video hit several million views and growth took off.

When to launch by: launch and build the account before March 2027. GlovioAI reached 75k followers in a month, which shows the format is easy to copy. Once 3D mannequin tutorials become a common video type on TikTok, new accounts won't get the recommendation boost. The New Year fitness season (January) is the best launch window.

Signs you've got it wrong: ① BoxMaxxing's Android daily installs fall to a few hundred in December and its new TikToks stay under a million views, which would mean the traffic comes fast and leaves fast, and can't be kept; ② batches of one-star reviews on iOS and Play about paying and then not being able to use the app, which would mean the hard paywall plus cheap global traffic can't sustain conversion; ③ if you search TikTok for your chosen sport and two or three 3D mannequin accounts are already posting, you're too late, so pick another sport; ④ conversely, if GlovioAI reaches tens of thousands of installs within a month or two of launch, the approach works for anyone and you should move faster.

---

Data and sources: Google Play store pages, install counts and charts (11-05 to 11-06; 8 to October daily installs include third-party estimates); App Store US store page and reviews (11-06); iOS country charts (11-06); TikTok @boxmaxxing.app profile, video views and comments on the 9-20 video (11-06); Instagram follower counts from search result summaries; MetaTensor's website and Australian business registry; competitor data from their store pages and heavybag.pro's AI boxing app reviews. Installs per thousand views, the rating prompt and whether JabMaxx is copying it are all inferences. Revenue amounts and paid acquisition spend are not visible from outside.