How Gravl grows

Posted a Reddit thread on how he built the app (then called Gains AI), sharing the technical specs. The audience was developers who like tech and may find the gym intimidating. The app was free at this point; the flood of bug reports and feature requests was the signal to add a subscription and start ads.

Ran paid ads immediately after adding the subscription model, having first validated usage via the free Reddit launch. Spend is mostly Meta and TikTok, with a little Google and Apple Search Ads. Ad spend runs at about a third of revenue (excluding creative production). Advises validating that people will pay before spending on ads.

Both founders speak Spanish, so they translated the app into Spanish and ran ads in South America, spending less than $50/day, instead of starting in the expensive, competitive US market.

Dropped a me-too Hevy/Strong-style workout tracker that added no value. After a friend showed him Fitbod, he used it and researched it, finding its generated workouts weird and sometimes dangerous. Built 'Fitbod UI/UX with an actual proper workout engine' that accounts for equipment, weekly goal, gym frequency, consistency, gender, weight and age. MVP took 2-3 months.

Right after the plan summary, users hit a hard paywall before signing in for the first time; no free access to the product.

In-app support chat available 24/7 with help articles and messaging; replies come from an actual person, not an AI.

Onboarding collects name, reason for using the app, experience level (advanced users get one-rep-max questions), goals, training frequency, split, excluded/focus muscles and gym equipment profile. AI is labelled throughout ('make sure that you are showing the users that this is an AI app, because that sells now'). A generating screen shows the custom workout being built, then a plan summary before the paywall.

Start small: pay creators as little as $50 per piece of content, use AI tools to generate videos and CapCut to edit your own. UGC is most of what works; AI video is cheap and easy to test. The hard part is volume, not making one ad. Use the public Meta Ad Library to see competitors' ads and where they put money, and copy what works.