The Cure.fit app used AI, ML, and data analytics to provide multiple health-oriented services in one app.

The Cure.fit app used AI, ML, and data analytics to provide multiple health-oriented services in one app. Source: Shutterstock

The AI-powered virtual fitness app taking India by storm

  • Cure.fit is India’s most popular health and fitness application
  • The company leans heavily on big data and AI to drive user experience and growth

Fitness and lifestyle apps have received a shot in the arm in 2020. With large chunks of the world’s population confined to their homes for months these platforms have offered a way for people to stay fit and healthy, physically and mentally in difficult and limiting circumstances. Lockdowns were especially strict in India, and this presented a growth opportunity for homegrown health and fitness startup Cure.fit to scale its tech-driven platform.

Launched around five years ago by the creators of e-commerce upstart Flipkart, Cure.fit is a Bangalore-based integrated app that provides its users to a suite of innovative, virtual classes for health and wellness. Physical fitness is delivered via Cult.fit, that provides workout classes in various formats like dance, yoga, and strength; mental wellbeing is under Mind.fit; Eat.fit is for healthy eating and foods; while Care.fit is a primary healthcare and specialist diagnosis platform.

According to Ankit Gupta, the Engineering lead at Cure.Fit, data analytics plays a crucial role in all the major decisions made around the app, its partnerships, and its growth. The app’s backend sports an intricate collection of over 100 subsystems or “microservices”, with each one modified to address a particular problem.

These microservices are improved by harnessing powerful machine learning models that have been trained on “millions and billions of data points” to achieve better results over time. From personalized programs, to class scheduling and allocation, all applications “are backed by algorithms that automatically become better over time,” Gupta told ETCIO.

The platform engages AI to tailor fitness plans to individual users, as the amount of data crunched would be impossible for human talent to perform.

“We use this in our meal and workout recommendations […] these solutions look at past behavior of users, compute preferences and infer health goals. Keeping all this in mind, the AI marries it with the menu or the class schedule for that day and comes up with the best recommendations based on availability,” explained Gupta.

The live-streaming fitness classes on Live.fit also features the proprietary ‘Energy Meter’ feature, that captures explosive movements when the camera is enabled using advanced AI-driven computer vision tech, to assign an energy score that can be compared with other participants in the virtual class.

“This feature is camera-enabled and, after getting the user’s consent, uses sophisticated computer vision techniques to infer what movement the user is doing in the scene,” said Gupta. “It tracks every joint in the customer’s body and approximates the amount of energy spent. Based on this, it creates a real-time leaderboard and automatically prepares a fitness report that is shown to the user at the end of the workout session.”

Cure.fit has been resorting to containerized, open-source infrastructure to maximize utilization efficiencies, reduce wastage, and control costs. “We solve pretty complex computer science, computer vision, and AI problems. This requires the infrastructure to be capable enough to handle any amount of data, traffic, and lots of computation,” elaborated Gupta.

“This also means using good reliable cloud solutions and marrying it with our optimizations that make the most sense for our business,” he said. “At the same time, we don’t take things for granted and have a lot of fallbacks, checks and balances in place.”

A unique case though it is, Cure.fit’s approach to leveraging the full power of AI across its product suite represents just how ingrained the technology is becoming, and how much of a competitive differentiator it can be – whether its potential manifests in consumer-facing products and services, or in the backend nuts-and-bolts.