Can AI help people lose weight?

As weight loss drugs like Ozempic have become massively popular in the west, a Singapore-based startup is attempting to provide a slower approach, centered around behavioural changes. Co-founded by Philip Man, Welling AI hopes to offer a digital alternative to the medicalisation of weight loss, using artificial intelligence to help users adopt healthier habits that fit into their real lives.

In the weight loss industry, Ozempic, a GLP-1 receptor agonist originally developed to treat type 2 diabetes, has quickly become the poster child of the weight loss drug industry. By working with the body to manage blood sugar levels, it also suppresses appetite and assists with rapid weight loss.

Correspondingly, it has found wide popularity beyond its initial purpose. In the US alone, the drug accounted for 5.4% of all prescriptions last year. Globally, sales of GLP-1 drugs surpassed US$100 billion (around SG$128 billion) in 2024, and analysts expect this number to increase to US$150 billion (around SG193 billion) by the early 2030s.

But while the enthusiasm is high, these medications come with trade-offs. They are expensive, require continued use, and often lead to weight regain once patients stop taking them. At least, unless meaningful lifestyle changes are made alongside taking these drugs.

Welling AI

It’s clear that a lot of people are concerned with weight, for reasons related both to health and appearance. As a result, Man founded Welling AI to challenge this medicalisation of weight loss. 

The app provides users with a virtual nutrition coach powered by AI. Rather than asking users to obsessively count calories or follow rigid plans, Welling AI lets users log meals through photos, voice, or text, and receive personalised feedback based on their input and goals.

A key strength of the platform lies in its localisation. Welling AI is trained on Asian food data, which allows it to differentiate between cultural eating contexts — for example, the difference between consuming meat at a Korean barbeque, hotpot, or a Western steakhouse. It also learns from individual users over time, adjusting to how each person defines terms like “a portion” or “a bowl,” which adds a valuable layer of nuance to its estimates.

The difficulties around weight

But despite its thoughtful design, the app still falls into similar struggles as its digital counterparts, especially for users navigating Singapore’s hawker centres. Hawker centre dishes are complex, non-standardised, and often unique to the stall or cook preparing them.

Estimating the calories in a plate of char kway teow, which might include varying combinations of noodles, the presence or absence of ingredients such as cockles and Chinese sausage, or different oil levels, is inherently difficult. For many users, the appeal of snapping a photo and receiving accurate nutritional feedback may quickly wear thin if the results feel too generic or inconsistent.

Another question surrounds how the app handles goal-setting. One of the persistent issues in the weight loss industry is that many users enter with unrealistic expectations, shaped by media imagery and past diet failures. While Welling AI claims to offer one-to-one coaching, it’s unclear how well the system navigates the complex emotional terrain of self-image, body expectations, and what it means to pursue “realistic” change.

Still, the company’s focus on long-term behaviour change and not instant quick fix weight loss, is a positive sign. Its focus on habit-building and cultural context suggests a more empathetic vision for digital health tools. 

As Welling AI continues to expand its data analysis capabilities, it could serve not just as a weight management tool, but as a case study in how technology can nudge individuals toward better health without the side effects.