The Practical Path to AI ROI in Fitness

Karl Foster, Head of AI at Sport Alliance - the umbrella company behind PerfectGym - is responsible for the software’s innovative AI developments, empowering gym operators worldwide. From developing cutting-edge AI solutions to shaping the way clubs engage with their members, his work touches everything from strategy to execution. In this interview, Karl opens up about his journey, the challenges fitness clubs face, and the ideas that inspire him - offering a rare behind the scenes of AI expertise.

 

 

 

Who is Karl Foster? A profile of the Sport Alliance’s AI expert

“My name is Karl Foster. I’m the Head of AI at Sport Alliance, where I lead the development of AI systems designed to future-proof the businesses of fitness operators across our PerfectGym platform.

My background has always been on the operator side of the fitness industry. I spent over a decade building automation systems, business intelligence platforms, and machine learning models for large-scale gym operations, most notably as CTO at GymNation in the UAE & KSA. There, my team and I pioneered genuine firsts for the sector, including the world’s first AI voice sales agent in fitness, agentic AI systems for member engagement, and proprietary fintech machine-learning-based solutions that delivered significant bottom-line impact.

After that, I founded my own AI consultancy, advising organisations on the strategic execution of AI. That work gave me a distinctive viewpoint across multiple companies, which is now complemented by the academic lens of my MBA research, where I’m specifically studying how businesses achieve – and fail to achieve – ROI from AI. So I’ve been fortunate to approach this space from three perspectives: as an operator who’s built and deployed AI systems, as a researcher studying the patterns behind success and failure, and now as a vendor building the next generation of AI tools for the fitness industry.”

The Roots of Expertise: How long have you been working in the fitness industry?

“Over 15 years. My very first job was as a gym instructor in the UK. I’ve since worked my way through every level of the industry – from the gym floor to the C-suite – across the UK, the UAE, and Saudi Arabia. That journey is actually what gives me such strong conviction about where AI can and cannot add value. I’ve been the person on the floor dealing with members, the person in the boardroom looking at P&Ls, and now the person building the technology. That end-to-end perspective is rare, and it matters enormously when you’re designing AI systems for this sector.”

From Pioneer to Head of AI: Karl, how long have you been involved in AI innovation?

“Around six years. My first work in machine learning actually had nothing to do with fitness – I was building trading algorithms for digital assets, trying to model the psychological and ledger patterns that move prices. 

When the generative AI boom arrived, I’d already been working with classification models, churn-prediction pipelines, and anomaly detection in production. So rather than starting from zero, I was able to dive headfirst into generative AI in 2023.”

What is your experience with AI?

“Broad and hands-on. My work included the implementation of the world's first agentic AI system in the fitness industry. Within this framework, for example, a voice AI sales agent was launched. Furthermore, a proprietary Smart Billing Engine based on advanced machine learning was built, delivering significant value to the company's bottom line. Object detection on CCTV, member segmentation models, churn prediction pipelines, and multi-agent orchestration systems were also successfully rolled out.

Across my career, I’ve implemented AI across the full member lifecycle – acquisition, onboarding, engagement, retention, share of wallet, collections, churn prediction, and win-back. I’ve worked with everything from traditional ML to generative AI to multi-agent agentic architectures. And now, having consulted for multiple organisations and studied the academic literature extensively, I’ve seen what works, what doesn’t, and – critically – why.”

Which challenges in everyday gym life can be solved particularly well with AI? Which tasks can be handed over to AI?

“I’d reframe the question slightly. It’s not about handing tasks over to AI. It’s about enabling AI to act as a force multiplier for your team. The literature is clear on this, and my own experience reinforces it: the greatest value from AI in our sector comes from augmenting human capability in customer engagement, not replacing it.

“The question isn’t which tasks you hand over to AI. It’s how you turn one team member into three.” - Karl Foster

The most immediate opportunity is in member services. Handling routine enquiries that don’t require human judgement – processing a membership freeze where your terms permit it, answering FAQs about opening hours or class schedules, updating payment details, these procedures should be instant and frictionless. Most gyms today either handle these manually or don’t handle them at all outside of business hours. Both are missed opportunities.

The second area is sales. AI can assist throughout the acquisition funnel – managing free trial bookings, answering pricing questions, overcoming common objections, and qualifying leads. This doesn’t replace your sales team. It guarantees no lead goes cold because someone didn’t pick up the phone at 9pm on a Wednesday.

But the real transformation – and this is where we’re focusing our energy at Sport Alliance – is in proactive engagement. Rather than waiting for a member to get in touch, AI takes the initiative. Imagine identifying a member whose visit frequency has dropped by 40% over the past three weeks, whose contract renewal is approaching, and whose profile suggests they’d be receptive to a personalised offer. The AI starts that conversation at the optimal moment, through the right channel, with the right context. That’s not a chatbot. That’s an intelligent engagement engine.

To summarise: Proactive hyper-personalised communication at scale, enabling members to get what they need faster with less friction, and supporting your team as a force multiplier – assisting them, not replacing them.”

Where can studios save costs and time with AI?

“Respectfully, I think this is the wrong framing, and it leads many organisations astray. The research repeatedly shows that companies that approach AI primarily through a cost-reduction lens tend to underperform those pursuing value creation.

That said, the productivity gains are meaningful. My MBA research examines what the literature calls the “J-curve effect” of AI ROI – a pattern where notable upfront investment precedes a period of negligible or negative returns before benefits accelerate. The research shows meaningful returns typically materialise at 18 to 36 months. The problem? Most organisations are abandoning their AI initiatives at month 12, right before the curve inflects upwards. 

When we think about ROI, there are two dimensions. Hard ROI includes increases in conversion rates, decreases in churn rates, increased referral generation, increased upgrade penetration, and increased revenue per member. These are directly measurable and attributable. Soft ROI includes staff productivity gains, morale improvement, ticket deflection rates, resolution times, member sentiment scores, and competitive positioning. These are harder to quantify but no less real.

The opportunity cost of not implementing AI is an increasingly more important calculation. My view is that 2026 and 2027 will create the largest gap we’ve ever seen between technology leaders and laggards in our industry. The organisations that implemented early, correctly, and stayed the course are now stabilising their AI engines and compounding value. Those still on the sidelines will find the gap increasingly difficult to close.”

How does AI improve the experience for fitness club members?

“This is where I get quite passionate, because I think many people in our industry misunderstand the role of AI in member experience. The goal is not to make the gym experience more digital. It’s the opposite. AI should remove the digital friction so members can maximise the time they spend doing what they actually came for: exercising, socialising, and engaging with the physical environment.

Think about it from the member’s perspective. They don’t want to send an email and wait three days for a response about a billing query. They don’t want to call the gym during their lunch break to freeze a membership. They don’t want to wonder whether anyone has noticed they haven’t visited in a month. AI solves all of this invisibly. The member gets speed, availability, and the feeling that their gym actually knows and cares about them. That last point is critical, especially for high-volume, low-cost operators or large multi-site businesses where the human bandwidth simply doesn’t exist to maintain personal relationships with every member.

“AI should handle everything before your members walk through the door and after they leave. When they’re in the gym, the experience should be entirely interpersonal.” - Karl Foster

That’s the key principle. AI enhances the connection between the member and the gym without ever interfering with the social and human dynamics of being in a brick-and-mortar facility. It’s the invisible layer that removes friction from the relationship, so that when members are actually in the gym, the experience is entirely about people.”

What role does AI play in supporting employees’ work in fitness clubs?

“There are two dimensions to this, and most organisations only see one of them. The first is what the industry calls “shadow AI” – individuals within your organisation using AI tools on their own initiative, often without any governance or standardisation. Your reception staff might be using ChatGPT to draft emails. Your marketing person might be generating social media content with an AI tool you’ve never heard of. This is happening in virtually every organisation right now, and it provides both an opportunity and a risk. The opportunity is that your team is already demonstrating AI-powered productivity. The risks include inconsistent quality, AI slop, data security risks, and the absence of institutional learning from these efforts. The second, and far more powerful, dimension is institutionalised AI – where you, as an organisation, deliberately deploy AI tools that make your team’s work better, more productive, and more enjoyable. This goes beyond generic productivity tools. In the context of what we’re building at Sport Alliance, it means your staff no longer need to sift through noise to figure out which members need attention. The AI does the heavy lifting – monitoring behaviour patterns, identifying optimal intervention stages, drafting personalised communications – and only engages your team when human involvement will genuinely add value. This frees up your staff for the work that actually requires a human touch: complex complaints, personal training consultations, community building, and relationship-driven work.”

What advice would you give to studios that are dealing with AI for the first time? Where should a studio realistically start?

“First, work with a specialist vendor. Building internally is neither viable nor sensible for the vast majority of fitness operators. Hiring a CTO, technical leads, engineers, maintaining systems, committing that level of resource – it’s not just expensive upfront, it comes with enormous hidden costs. The literature is quite clear: ROI is more likely to succeed, and the volume of returns is substantially higher when working with a specialist partner. My own experience building from scratch gave me a deep appreciation for just how much organisational learning, iteration, and frankly, failure is involved before you reach production-grade AI. That’s a journey most operators don’t need to take themselves.

Second, understand the two fundamental barriers. When working with a third-party AI solution, there are two challenges that determine whether value can flow. The first is connectivity and integration – third-party solutions are not native by design. They live outside your ecosystem, connect via an API layer, pull your data out, and push it back. This creates fragility. The second is the reactive versus proactive problem – most AI solutions can only act when the end user decides to engage. They’re sitting idle 99% of the time, waiting for someone to start a conversation.

Both of these barriers lead to what the literature and my own peer feedback regularly describe as “decay.” Six months after implementation, you change your membership tiers or update your booking flow, and suddenly the AI breaks. The value of an AI agent lies in its ability to take actions, process workflows, and complete tasks with the right knowledge, context, and capabilities. But when the system is bolted on rather than built in, the risk of poor performance increases over time as your organisation evolves, while the AI does not.

“Bolted-on AI decays over time. Native AI evolves. This is the difference between a system that breaks in six months and one that compounds value over years.” - Karl Foster

Third, start small and measure everything. Identify a specific use case – member services, sales, or retention. Define how you’ll track impact before you deploy. Understand how the AI can add value in that specific context. Then go step by step, testing and learning as you go. What works in one club might not work in another. Learning builds over time, and that compounding is what gives you an increasingly strong advantage over your competitors.”

How do you think AI will develop in the fitness market in the coming years?

“One thing is beyond question: AI is here to stay. But I don’t believe it will replace humans in any meaningful way in our industry. What it will do is take an increasingly active role in the management of memberships, sales, retention, and collections.

We’ll see a fundamental shift from reactive to proactive AI agents. Rather than waiting for a member to ask a question, AI will take an active role across every stage of the member journey – from the first website visit through onboarding, ongoing engagement, and even when a member decides to leave. The AI will understand the entire context of each member’s relationship with the gym at any point in time, and that situational understanding is where AI truly excels compared to any human-managed process at scale.

I believe the majority of the sales process, onboarding, member engagement, and even offboarding will be managed primarily by AI within the next three to five years. Humans will remain the first point of contact for situations that actually require care and nuanced judgment – complaints, sensitive personal situations, and complex service recovery scenarios.

Beyond communication, AI will play a growing role in operational optimisation – price elasticity studies, class scheduling, demand forecasting, and resource deployment. It will add to programme design, such as personalised workout plans and training recommendations. However, I feel strongly that AI should not interfere with the exercise experience itself. The gym floor is a social environment. It’s a community. That human dynamic is sacred to our industry, and it’s the one thing AI should never attempt to replicate.”

What are you currently working on at Sport Alliance?

At Sport Alliance, we’ve been obsessing over two specific things:

The first is connectivity and integration. Third-party AI solutions are not native by design. They live outside your ecosystem, connecting via an API layer to pull your data out and push it back. This creates inherent fragility, and it’s the root cause of the “decay” problem that surfaces consistently in the literature and in peer feedback. 

The second is proactivity. Most AI solutions on the market today are purely reactive. They sit idle until a member initiates a conversation. That means they’re only harnessing a fraction of the value that could be unlocked through proactive engagement – reaching out to the right member, at the right time, through the right channel, with the right context.

So what happens when you remove both of those barriers? In my opinion, you get AI systems that genuinely future-proof your business. 

And this is what we’ve been building: PerfectAI – native AI that can access your system (PerfectGym) the same way your staff can, without limitation. When used together with our proactive engagement engine, the AI works autonomously, learning continuously from a single source of truth. 

We’ve already entered an alpha-phase pilot, and the results are promising. We’ll be sharing more information soon about our release timeline. For now, I’d simply say: the two biggest technical barriers to AI value in fitness are being solved, and we’re the ones solving them.”

At which events or conferences can people meet you and speak with you in person?

“I’ll be presenting a keynote at FIBO 2026 in Cologne on April 17th at the Women’s Leadership Forum, titled “Achieving ROI in AI in the Health and Fitness Industry.” I’ll be unpacking the research behind the AI hype cycle, the J-curve effect on ROI, and what leaders in our sector need to be doing right now to stay ahead. I’ll also be available on the Sport Alliance stand (CONC32 + CONC36) throughout the event, where we’ll be running live demos of our AI innovations.”

Image