George Lampropoulos - Meet the ex-wrestler making $41k/mo with AI apps

Non-technical founder used Rork (AI app builder) to build an idea he had for a while: upload a video of your martial arts training and get told what you did right and wrong. He and a friend posted about it on social, which made ~$1K at near-zero cost.

FightAI covered any martial art. Users repeatedly asked 'Does it do wrestling?', so he built a wrestling-only app (Wrestle AI) for a niche he knew from wrestling through high school. Because no other wrestling app existed, wrestling audiences had never been hit with app sponsorships, unlike fitness, where Cal AI-style subtle integrations are fatigued. Then repeated the same product for boxing (Brawler) and is working on an MMA version.

George's first ever expense was a virtual assistant running the app's Instagram DMs, and he says it should be everyone's first hire. Before a sponsorship call with a smaller creator for FightAI, he noticed the creator's profile showed 'Followed by Cayden Henschel'. He cancelled the call and had the VA ask that creator to DM Cayden; Cayden replied, then answered George's email, they swapped numbers and became partners.

Partnered 50/50 with Cayden Henschel, the biggest wrestling creator. No posting minimums or vesting were set (George says he got lucky and recommends setting minimum posts and vesting). Cayden posts about the app (e.g. announcing the rebuilt product, stitching a coach calling out moves with 'this app will call out the moves for you'), and recorded all the combos in his own voice; the app randomly mixes them for drills.

Set up an App Store pre-order and had Cayden hype it to his audience so everyone pre-ordered; downloads landed together on launch day.

Sponsored other wrestling influencers on top of Cayden: roughly 10 at a time, never 100. As results came in, cut the roster and doubled down on the ones that worked. Since there are few wrestling creators, they try to sign as many as possible to own mindshare in season. Recommended for small budgets in unsaturated niches; harder to manage at scale.

Started paid ads for Wrestle AI the week before the episode. Pick the best videos, not always the top organic ones, because some organic winners take too long to reach the CTA (get it in within ~15 seconds). Main criterion is clarity: show exactly what the product does (e.g. uploading a match and seeing what you did wrong) in a way that isn't boring. Then a few strong creatives and scale spend.

Hook is a third-party clip (e.g. a convoluted AI-looking workout); it cuts to the creator saying 'This is just ridiculous. If you just download this app, it'll give you the exact workouts you need'. In the top ad the hook runs ~9 seconds before he talks; aim to land the CTA within 15 seconds. Same format on Wrestle AI: Cayden stitched a coach calling out moves. Made by AbMax's one in-house creator (Calvin); George and Lino run them.

Ad ideas that wouldn't work organically go to paid. Example: Cayden ranks the best wrestling apps (Notes app for a practice journal, MyFitnessPal for fueling) and ends 'if you want all of these, I would use Wrestle AI, my favorite wrestling app'.

The whole ad just shows how the Brawler app works (pad-work mode with targets coming at you), a raw screen recording with no thought put into editing or the CTA.

Went all out several times on 'really sick looking' visuals for ads.

Whenever an ad works, create a bunch of variations. Tried scaling spend fast; CAC rose until results evened out, so they increase spend slowly. Bottleneck is creative volume: they think they need ~10x current creative output to scale spend fast.

To raise creative volume, they recruit more UGC ad creators and have them copy what their main creator (Calvin) does, then run their videos as ads.

Calculate the break-even CPA (e.g. $40 per purchase on Brawler) and judge ads purely on whether CPA moves up or down. Works on AbMax because it runs only paid ads; the new clipping campaign will muddy it. Hired Cal AI's back-end/ads person as a contractor to fix attribution. Onboarding also asks 'How did you hear about us?' to help track attribution.

Paid a boxing creator with ~1M followers $4,500 for three videos as a mindshare play, aiming to recoup only 25-50%. Idea: boxers see someone they trust using the app, then get hit with the paid ads, which should lift all paid metrics.

For the boxing app, look for creators making content for people who want to get good at boxing (can box well, explain clearly, tell a story), not highlight accounts for people who like watching boxing. Few such creators exist, which limits volume.

Clipping campaign of educational slideshows on how to get abs ('you have to do this, this and that; if you want it all in one place, get AbMax'), an evergreen format that can be replicated endlessly. At ~$0.30 CPM it can lose money, because the goal is ~10M views so paid-ad CAC falls. Based on Rork Max going viral on Twitter and Facebook CAC dropping for a month.

Same format in all three apps: educate (show the main feature, e.g. film analysis that circles mistakes, drills, stance mode, weight-class fueling), social proof ('Join over 100,000 wrestlers' with creator footage), 'How did you hear about us?', then personalize: style, go-to attack, best position, pain-point questions ('what's costing you most matches', 'hardest part about making weight'), training frequency, height/weight plan, optional match-schedule upload, a stat screen, projected-growth graph, optional match upload. Then the paywall ($9.99/month or $59/year on Wrestle AI, ~50/50 mix) showing a generic example breakdown as a taste of the magic moment.

Match analysis is only used after each match, so they built a solid calorie tracker for wrestlers making weight. Users can upload their season match schedule and get reminders like 'your match is in 4 days and you're 5 lb over; here's how to make weight'.

Wrestle AI is the only one of the three apps with a free trial; Brawler and AbMax sell straight to purchase. They are now testing taking the trial out of Wrestle AI to see how it converts.

Plan to run heavy Christmas ads aimed at parents buying the app for their high-school kids. Still working out how to sell prepaid access without breaking Apple's rules (offer codes can be given out but not sold); reselling Apple gift cards is an option but risks losing the buyer.

Rolled out a Twitter-style feed that opens a TikTok-style For You page on video tap, full of wrestling tips. Cold start: every sponsored creator must post inside the app whenever they post on social. Ranking is a simple point system (likes, comments) heavily weighted to new content because a paid app has few posters. Goal: 'come for the tool, stay for the community', since users only analyze a match after each match.

Inspired by Ladder, paid a coach who trains many pro fighters to record boxing instructionals. Users go through them and the live vision model checks they actually hold a proper boxing stance before unlocking the next lesson.

Frequent notifications reverse-engineered from Duolingo's and applied to wrestling ('Hey, you didn't track your calories. What's going on?', 'Lock in'). Tied to behavior: tracked calories yesterday -> 'track again'; went on a run -> 'beat your record'. Users who finish onboarding without paying, and lapsed users, get 'Don't give up on yourself'-style copy.

On the third app open, a pop-up asks 'Do you want to talk to one of the founders?'; users enter their number and the founders call random sign-ups, aiming for about an hour a week.

For complicated features, play a screen recording or a creator explaining it the first time the user taps it, then never again (e.g. the punch-counter 'compete' challenge, then 'I got it. Let's go'). Creators record these for free because they like being in the app. Prompted by a sponsored creator who filmed a scripted video using the wrong feature because the pad-work button wasn't clear; they track usage in PostHog.

Ladder is completely free for the first 7 days and shows no paywall at all, so it relies on getting users to return many times; George calls it an unusual strategy.