Recently, at the Ignite conference, Khosla Ventures-backed Indian health and fitness startup Healthify (formerly known as Healthifyme) introduced Ria 2.0, a generative AI-powered personal health coach with multimodal, multilingual conversational capabilities.
The chatbot health advice adjusts to individual lifestyles and goals by considering factors like schedule, meals (captured via photos), blood sugar levels, activity, and dietary preferences.
Healthify’s ‘Snap’ feature, introduced this year, revolutionises nutrition tracking by allowing users to photograph meals for instant identification of various food types and calculation of nutritional content, including calorie count. The company claims that it is 40% more accurate than the original version, recognising over one million global foods.
Two other announcements were also made at the event. Firstly, the fintech company has entered a commercial partnership with food commerce platform Swiggy to allow users to order meals according to Ria’s smart dietary recommendations through the Healthify app, facilitating adherence to a healthy diet.
Secondly, Healthify’s “Coach Co-Pilot”, launched in 2023, which combines AI with human coaching, was validated by a Stanford study for a 70% increase in weight-loss effectiveness.
Inside Healthify’s Generative AI Game
“Our approach involves using multiple foundational models to contextualise information based on specific tasks and not relying on one for all,” Tushar Vashisht, cofounder and chief executive officer told AIM during the event.
Healthify’s team uses solution models based on diverse statistical models, including AWS, Meta, OpenAI, and open-source options, with a focus on domain knowledge for new insights.
“Leveraging an extensive dataset of over 350 million messages and diet plans, their approach involves assembling different models tailored to specific outputs, such as enhancing conversational engagement in voice interactions with user data analysis,” Abhijit Khasnis, vice president – technology, told AIM. When standard foundation models fall short, the team strategically leverages specific use cases to interact more effectively with real-world data.
Another important point is that the company places a strong emphasis on user control and confidentiality in handling health data. Customers can delete their data and decide whether to share it with their coach. “We never participate in data partnerships or selling user data, aligning with regulatory and ethical considerations and customers have full control over it,” noted Vashisht.
Growing with AI Through Years
According to the company, one significant aspect of its journey has been the shift from older AI systems like the original Ria (launched in 2019) to the new generative AI systems.
“Our initial AI could only handle a limited scope of queries and data points. However, with advancements, we’ve expanded our data vectors to include a wide array of inputs like calendar, geolocation, heart rate, nutrition, and fitness, greatly enhancing the comprehensiveness of our service,” he added.
Agreeing with Vashisht, Anjan Bhojarajan, chief business officer recounted that another turning point came when the team first considered monetisation for their AI services back in 2022 when ChatGPT was born.
“Initially, we thought it would only improve conversations, but it quickly became apparent that this new mode of communication would improve app interfaces, leading to the development of a new app with a chat messaging system, marking a shift from traditional input methods to a more unified and interactive interface,” Bhojarajan added.
Currently, Healthify is preparing for international expansion, and this financial year is set to be the most profitable yet, despite restructuring efforts and layoffs last fiscal year, according to Vashisht.
“For FY24, we are expecting double-digit growth, crossing $30 million in revenues and approaching a $40 million run rate while maintaining lower costs than ever before,” he concluded.