Every large fitness software vendor is now selling the same idea: move your whole club onto their platform and let their AI run it for you. At UPTIVO we think this is the last chapter of the pre-AI world, and we built our platform the other way, open at the level of devices, data and AI agents, before AI has even made the question urgent.
Recent months showed how far the closed model has gone. Playlist, the parent of Mindbody, ClassPass and EGYM, has made three AI launches for operators since July, starting with a 24/7 AI front desk.
On September 22 it added AI Insights, an analytics layer it describes as "powered by one of the world's largest fitness and wellness operating systems." They are useful, but built so that the AI sees only what lives inside the vendor's own suite.
Closed suites are putting AI on top of an old lock-in
If one company holds bookings, payments, equipment data and the member app, its models can reason across all of it, and it becomes ever harder for clubs to leave.
And the fight over data has already started. In June 2025 Salesforce changed Slack's API terms to stop third-party apps from storing or indexing messages long term, a move that hit AI companies such as Glean directly.
An AI that sees only part of the data will give imprecise answers
It is common for an independent club to run bookings and billing in one system, door access in another, heart rate training in a third, marketing automation in a CRM and accounting somewhere else, because each was chosen as the best tool for its job.
An AI inside the booking platform can see that a member still books two classes a week, but it cannot see that her heart rate sessions have been getting shorter for a month, or that she stopped opening the club's emails in August, because those facts sit in systems it was never given.
The club's operating system will be an AI chosen by the club, not by the vendor
This is not only our view. Microsoft CEO Satya Nadella has described business applications as databases with business logic on top and predicted that the logic will migrate to AI agents, and OpenAI co-founder Andrej Karpathy has long described an "LLM operating system" in which the model coordinates tools and data the way an operating system coordinates hardware.
In a gym, that means the club will have its own orchestrator agent, ChatGPT, Claude, Grok or whatever comes next, coordinating more specialized agents across every system the club runs: the bank account and the reports for the accountant, bookings, door access, the cameras, staff and training. The club chooses that agent, not the vendor.
The plumbing already exists. Agents are getting better at reading a well-documented API directly, and the Model Context Protocol, the open standard that began at Anthropic and is now backed by the major AI companies, gives an owner a simple way to connect them with a login and decide what they are allowed to do.
In that world a vendor's feature list matters far less than its openness. Every vendor will have to send and receive data in real time and let the club's agent read it and act on it, and a closed suite becomes a wall the agent has to work around, while an open platform becomes part of the club's intelligence.
UPTIVO gives AI senses and hands inside the gym
However capable it is, an agent is blind and handless inside a gym. It does not know who is working hard on the floor right now, it cannot switch on the class screen and it cannot open a turnstile, and that is exactly the piece UPTIVO has been building for years.
Openness starts with the members, who train with the wearables they already own, Apple Watch, Garmin, Polar, Whoop or any standard Bluetooth strap, so the club never locks them into a single sensor brand.
It continues with the data. UPTIVO has a REST API with an OpenAPI 3 specification and webhooks for real-time club events, publicly documented at developers.uptivo.fit, where the specification can be downloaded and read by a developer or an AI assistant alike. It is designed to work alongside the management software a club already runs, not to replace it.
The same API powers our Zapier app, with 25 triggers, 16 actions and 28 searches, which connects UPTIVO to more than 9,000 applications and CRMs without a line of code.
And it reaches the agents. UPTIVO is already listed in Zapier's MCP directory, so an owner can authorize an assistant like Claude to book a class, create or invite an athlete, or read an athlete's training telemetry today.
That makes a new kind of question possible. An owner can ask an assistant which members kept booking this month while their training intensity dropped, which requires both the schedule and the heart rate data, and the churn-risk score in UPTIVO DASH can feed the CRM the club already uses instead of sitting in yet another silo.
Open done properly still needs rules
Being open does not mean giving an agent the keys to everything. Bank data and camera footage in particular fall under the strictest privacy rules, so they need the tightest permissions.
The right model is to treat the AI the way you would treat a human employee, with a defined role and access only to what that role requires. For video surveillance, the analysis of footage stays with specialized systems: cloud platforms such as March Networks' MN CloudSight analyze video and raise events like a blocked exit or an intrusion into a restricted area, and the club's agent works on those alerts, not on the images.
The same goes for training data. NATE, our AI coaching, analyzes workouts, recovery and trends for each member inside a narrower authorization perimeter, with the explicit consent GDPR requires, and the club's agent can work on its summaries rather than on raw health data.
The way to start is read-only access, a short list of approved actions and a person who reviews what the agent proposes before anything reaches a member.
Five questions to put to any vendor before the next renewal
Ask whether the API is truly public, with documentation anyone can read, or whether using it takes an invitation or the vendor's approval, which leaves the vendor deciding who gets in. Ask too whether the system pushes events by webhook or makes you poll it, and whether API calls are metered.
Then ask whether the product can be reached by AI assistants today, through Zapier or MCP, and whether you can export a member's full history in a format another system can read. If most of the answers are no, the vendor is selling a wall with a shiny AI sign on it.
The big suites will keep adding closed AI inside their walls. But a club is a business like any other, and if it wants an assistant capable of seeing every member in full it should choose systems that open their data to it, the path UPTIVO chose from the start.
Sources
Playlist, Playlist Launches AI Insights, Giving Operators a Best-in-Class Intelligence Layer
Outlook Business, Microsoft CEO Satya Nadella Reveals How AI Agents Will Disrupt SaaS Models
CX Today, Salesforce Blocks Competitors from Using Slack Data: So What?