Most people are used to hearing about artificial Intelligence (AI) chatbots being used as personal companions, customer service reps, and so on. Some fitness-­focused individuals however are turning to AI chatbots as a ­personal digital coach to optimise their fitness journey.

Rizal Kamal started his fitness journey back in 2024 at 49 years old. At the time, he was 94kg and decided that he would lose weight to be a healthier and ­fitter version of himself.

He went on to join a fitness challenge that focused on losing fat, but it became apparent to him that while he knew how to lose weight, he didn’t know how to lose fat.

“I used AI to set my goals. It looked at my weight, my height, my age, my body fat, my skeletal muscle mass, and it was able to create a programme for me,” he says.

From that point on, Rizal ­utilised the pro version of ChatGPT as his personal trainer and meal planner to ­further empower his fat loss journey.

Asking the AI chatbot detailed questions – such as “what’s the principle behind losing fat?” and “what kind of food do I eat in order to reach my goals?” – gave Rizal a better understanding of his nutrition and how to approach it.

AI gave Rizal a better understanding of ­nutrition planning. — IZZRAFIQ ALIAS/The Star

And when it comes to his workouts, the AI gave him a ­better understanding of how many times he should work out in a week, balancing between weight training and playing ­pickleball, an activity that he ­particularly enjoys.

“It’s a continuous conversation, with ChatGPT, in order for me to fine-tune how I should design my fat loss journey,” he adds.

A similar physical evolution unfolded for Ben Trenchard. The 43-year-old decided to train for the Subang Jaya Half Marathon, turning to an AI-­powered fitness application called Runna to map out his training splits and analyse his physical performance.

His usage of AI, however, didn’t really start until he took the lead for AI implementation in his office, at which point Trenchard decided to explore its use beyond the office space.

“We got a Claude (Anthropic’s AI platform) plan, and everyone went a little bit AI crazy,” he recalls. After using it for work-­related tasks, he started using the AI tools to help him with personal goals, like building a better nutrition plan.

From there, he transitioned to using AI for his running workout once he realised that the Runna app was able to give him AI-­generated pre-workout guidance and post-workout analysis of his marathon training.

“I wanted to do more structured training, and I realised that with Runna, it actually uses AI to build the training plan as well. And just naturally, I’ll use it for nutrition planning and workout tracking,” Trenchard adds.

AI-optimised training

For Rizal and Trenchard, the initial jump into AI was only the first step. Both men realised they had to push their prompts beyond casual conversation and into the realm of hyper-personalised data inputs to further optimise their fitness journey.

Through Claude, Trenchard created his own custom workout tracker. — YAP CHEE HONG/The Star

“The most important thing when chatting with AI is giving it context and giving it as much information as possible. I went on a (body scan) machine that gave me about 20 or 30 different points of data, and I inserted it into the AI. The more data I insert into the AI, the more accurately it can guide me,” Rizal explains.

Giving such detailed information to AI has been the key to better understanding his body and optimising his fat loss journey, a process through which he has since shed 14kg.

“When you should eat, how much you should eat, what kind of food you should eat to control your blood sugar, AI can really help (when it comes) to the

minute detail,” Rizal adds.

To take his workouts to the next level, Trenchard wanted to eliminate as many distractions as possible when it comes to tracking his progress.

Rather than sifting through multiple workout tracking apps, he used AI coding to create his own hands-free, voice-to-text gym workout tracking tool.

“When you’re in the gym, you want to stay focused on what you’re doing. So, I used Claude Code to build something where I can just record a voice note for each workout set, put them all in a folder, and the AI will read those files and generate the workout summary for me,” Trenchard explains.

Where Trenchard deployed AI to construct an isolated tracking tool, Josh Foong took the ­technology to an athletic ­frontier.

The 33-year-old with years of bodybuilding experience progressively integrated AI over the past year as part of his cycling journey, one that sees him take on a gruelling cycling load of 200km to 300km a week, leaving no room for recovery errors.

“I’m using this platform called Intervals.icu,” Foong shared. “What it does is, it tracks every ride you upload – your heart rate, your speed, distance, elevation – and there’s a lot of meaningful data to it.”

For Foong, AI helps him improve his athletic performance. — AZLINA ABDULLAH/The Star

With all the data he collected from Intervals.icu, Foong used a connector to sync the data to Claude to give the AI platform constant access to his metrics, which allows the chatbot to ­better understand and help improve his athletic performance.

“What the integration with AI helps with is that it actually ­analyses your recent rides as well, telling you that, ‘Hey, in the last 20 minutes, I see a drift in your heart rate’, and it shows that your fatigue is coming in.

“So then, you start looking back and think, ‘Oh, you know, at that time I was really tired, but why?’ So you start analysing, thinking how you can ride ­better, and most importantly, ride smarter,” Foong adds.

Democratising fitness

The accessibility of AI allowed Rizal to be in constant conversation with his digital coach and meal planner, solving a “lack-of-time” issue that he believes he would otherwise have faced with a human specialist.

“You have 15 minutes, at most, you’ll have 30 minutes with a doctor. And when you explain, you tell them your symptoms, what you want, and then the ­doctor will give you advice. But it takes a lot more than 30 ­minutes for the doctor to fully understand you and give you the best advice.

“I talk to my AI over a span of months. And it’s a continuous process. I have my learning buddy right next to me that will guide me the whole way,” Rizal adds.

Dr Aiesha Asmadi, who has a background in sports medicine and orthopaedics as well as being a competitive bodybuilder herself, notes that the technology can be a great starting point, given how approachable AI can be.

“The good thing about AI is that it makes fitness advice more accessible. When you have AI, you feel less lost because it gives you a sense of control over your health and your nutrition.

“You can use it as a starting point in terms of knowledge,”

Dr Aiesha explains.

Flagging the risks of AI

Experts however caution against going to the other extreme and sharing detailed medical information with AI chatbots.

A recent study from the United Kingdom’s University of Oxford highlighted the risks of inaccurate and inconsistent medical advice that could present risks to users.

According to the research, ­people using AI for healthcare advice were often provided with a mix of good and poor recommendations, ­making it difficult to discern the best course of action.

"These findings highlight the difficulty of building AI systems that can genuinely support ­people in sensitive, high-stakes areas like health,” says Dr Rebecca Payne, the lead medical practitioner on the study.

Dr Aiesha says that AI can be a great educational tool, but it isn’t a treatment plan. — KAMARUL ARIFFIN/The Star

A New York Times report from August 2025 quoted Dr Danielle Bitterman, an assistant professor at Harvard Medical School and clinical lead for data science and artificial intelligence at Mass General Brigham: “Just because you’re providing all of this information to language models, doesn’t mean they’re effectively using that information in the same way that a physician would.”

As such, despite integrating Claude with Intervals.icu to provide sophisticated data insights, Foong doesn’t treat the chatbot as an infallible digital coach. “I argue with my AI a lot,” he quips. “Not everything that the AI proposed I would take as ­gospel from the first time. You have to keep refining it.”

During his usage of AI with his training, Foong admits that he has had moments when the ­chatbot was experiencing ­“hallucinations”, whereby the AI might propose a training plan or recovery time that might not make sense.

At these moments, he reiterates the need to challenge the AI ­rather than treating its outputs as absolute law. For Foong, the best defence against algorithmic error is not more data, but his own deep history of physical training.

“I think it helps that I’ve been bodybuilding for so many years,” he points out. “I think that’s a very big factor in ­understanding your own body because AI can’t tell you how you feel.”

It’s a level of scepticism that Trenchard shares with Foong when it comes to using AI as a repository of health and fitness information.

Even while using AI to ­manage his meal ­schedules or track his workouts, Trenchard understands that its capabilities are ­limited by the AI models he uses.

He points out that while AI is great when it comes to researching and putting information together, it doesn’t really do the thinking for the user.

“You have to remember, it’s just a language model,” he adds.

Another risk centres around data privacy – once people give chatbots detailed medical data, they may have very little control over how it is used, stored or handled, experts say.

A recent Stanford University study showed that six of the top US companies feed user inputs back into their models to improve functionalities, with only some giving users the ability to opt out.

The study’s lead author, Jennifer King, Privacy and Data Policy Fellow at the Stanford Institute for Human-Centered AI, says: “If you share sensitive information in a dialogue with ChatGPT, Gemini, or other frontier models, it may be collected and used for training, even if it’s in a separate file that you uploaded during the conversation.”

According to an FAQ by OpenAI, “we may use ­content submitted to ChatGPT and our other services for individuals to improve model performance”, your data may be shared with “a select group of trusted service providers that help us provide our services”, and other humans may view your content: “A ­limited number of authorised OpenAI personnel as well as trusted ­service providers... may access user content only as ­needed.”

The NYT report stated a few people had redacted their names and scrubbed metadata before sharing their medical records with chatbots, but Dr Rainu Kaushal – the chair of the department of population health sciences at Weill Cornell Medicine and NewYork-Presbyterian – says that medical data with enough details can still sometimes be linked back to individual users, even if there are no names attached.

A balanced approach

For Rizal, his 14kg weight loss journey taught him that artificial ­intelligence acts purely as an ­interactive mirror for a user’s personal clarity of purpose and honesty.

“The AI is ­essentially a ­reflection of me,” Rizal remarks thoughtfully. “Garbage in, garbage out. You can’t blame the AI because it can give you ­completely different answers on different days depending ­entirely on how you prompt it.”

Dr Aiesha, on the other hand, reiterates the fact that AI can be a great educator, but not a great ­doctor.

“It is excellent as an educational tool, for ­brainstorming, and for data collection,” she says. “But it is not a treatment plan.”

She emphasises that while the democratisation of information that AI provides can be a boon for public health ­literacy, it still needs to be used in a way that integrates advice from an expert.

“If you want to use AI, it’s best to use it in a way that complements the guidance you receive from a trained professional,” Dr Aiesha ­concludes.