AI in the gym: the most-cited case isn't independent

Search "AI chatbot for gyms" and you'll find almost the same case every time: GymNation, a gym chain in the UAE and Saudi Arabia, is said to have pushed lead conversion to 87%, against a 60-70% figure described as the industry standard, using a chatbot and a voice assistant built on LlamaIndex's technology. Before treating it as a benchmark for your own gym, it's worth checking where that number comes from.

The most-cited case comes from the company selling the technology

The 87% figure, alongside two others (a 75% automated tour-booking rate and a 65% actual tour-attendance rate, both against a stated 50% standard), is published by LlamaIndex, the company that supplies GymNation's technology for its two assistants, named Albus and Jenny AI. The exact same text, with the same numbers, appears on LlamaIndex's own blog and on GymNation's, which credits LlamaIndex as the source at the bottom of the post: these aren't two independent confirmations, it's a single piece of content republished twice. The "60-70%" and "50%" figures cited as industry standards have no source, study, or organization attached to them either: they appear as unreferenced claims within the vendor's own text.

There's another signal worth noting. Salesforce, another technology vendor used by the same chain for its CRM, publishes its own case study on GymNation that reports a 10% increase in sales conversion, not 20% as in LlamaIndex's text, and cites 110,000 members against the 90,000 cited elsewhere: different numbers for the same company, depending on who's publishing them. And even taking the best-case numbers at face value, they still describe a chain with 20 locations, over 90,000 members, and a dedicated AI development team in a highly digitized Gulf market: conditions far removed from those of a neighborhood gym in Italy.

The Italian market is already large, even without the most-cited number online

Searching for sector data in Italian often turns up a more recent, more striking cluster of figures (3.1 billion euros, over 7,500 centers, +7.5% members in a year) loosely attributed to a Deloitte report with EuropeActive. Opening the sources, though, that cluster doesn't hold up: the outlet that should confirm it, Il Sole 24 Ore, writes "over 5,000 clubs" in its own article on the topic, not 7,500, citing an earlier edition of the same report; the pages that repeat all four figures together copy each other without ever linking the original Deloitte report. The most solid figure I found, because it's dated and cites a statistical source, comes from an Istat and Unioncamere breakdown via TrueNumbers: in 2021 Italy had around 5,100 gyms, about 5.5 million members (8.3% of the population), and 2.3 billion euros in revenue. It's not the freshest data, but it's still a large market: large enough to make clear that opening a new fitness center today means competing in an already crowded sector, not a niche.

In Italy the real-world case doesn't have numbers yet, and that's worth saying

A real Italian example exists: FitUP, a chain that claims 81 clubs and 120,000 members, introduced a tool called Technogym Checkup starting in spring 2025, combining AI and sensors to assess body composition, strength, balance, mobility, and some cognitive parameters in a few minutes. The story, picked up by ANSA and other outlets between late March and July 2025 with the same text and the same quotes from the founder, has all the hallmarks of a company press release distributed multiple times, not a piece with its own journalistic verification: the 81 clubs and 120,000 members are FitUP's own figures, not confirmed by an independent source. No available source reports a measured result either, no extra sign-ups or converted leads attributed to the tool: it stands as a signal that Italy's larger chains are investing in tools like this, not as proof that they outperform human contact.

What it means if you run a gym or a fitness center

The Italian market is already large, and the bigger chains are investing in tools that combine AI with data collection: two good reasons to look at these tools carefully, not to buy one based on a foreign case promoted by the vendor that sells it. Before even considering a chatbot, what's worth measuring is your own funnel: how many inquiries come in each week, how many go unanswered, and how long it takes for the first response. It's the same starting point that applies in other sectors where a number publicized by a vendor needs checking before it's used as a sales argument: I wrote about a similar case in the auto industry, in AI in the repair shop, where a structurally similar figure, published by the same kind of source, didn't hold up to independent verification.

If you're evaluating an AI tool for your gym and want to start from your own funnel data rather than the most-cited case online, get in touch: I'll help you figure out what actually makes sense for you.