Earlier this month, executives from major tech giants like Google DeepMind, Microsoft, and Meta gathered in Bangalore to witness the launch of a groundbreaking product by Sarvam AI, a leading Indian artificial intelligence startup. Often dubbed “India’s OpenAI,” Sarvam AI unveiled voice-interactive software designed to transform how businesses in the world’s most populous country engage with customers. Unlike traditional text-based chatbots, Sarvam’s technology allows for spoken interactions in 10 native Indian languages, priced at an affordable rupee per minute to capture a vast market. Billionaire investor Vinod Khosla, who backs Sarvam AI, emphasized the potential reach of these voice bots, predicting they could connect with a billion people.
India’s tech scene has been striving to keep pace with the global AI boom since the introduction of ChatGPT nearly two years ago. However, progress has been hampered by the limited availability of data in many of the country’s languages. While urban dwellers may interact with AI in English, a significant portion of the population lacks the language skills to do so. This gap has led a growing number of startups to focus on voice bots that utilize local language data, aiming to broaden AI accessibility across India and potentially appeal to international users as well.
These startups are positioning India as a testing ground for the next wave of generative AI products, incorporating voice features to create more interactive and responsive services. This trend is already evident in various consumer and business applications throughout the country.
For instance, Samsung-backed Gnani AI facilitates millions of voice interactions daily for major Indian banks, insurers, and car manufacturers. CoRover AI provides voice bots in 14 Indian languages for entities like the state-owned railway corporation and regional police forces. Meanwhile, Haloocom Technologies offers a voice bot capable of conversing in five Indian languages, managing customer service inquiries and assisting in job candidate screenings.
“Voice is the most intuitive way to use technology,” said Ankush Sabharwal, co-founder and CEO of CoRover. The company’s Ask Disha voice bot, recently launched for India’s train booking service IRCTC, enables customers to book tickets and make payments entirely through voice commands. According to Sabharwal, India needs AI agents that can perform tasks, not just provide information.
Gnani AI also supports lenders in engaging with potential customers, gathering personal information, and assessing loan eligibility. Additionally, the startup collaborates with Tata Motors to collect customer feedback on new car models and market extended warranties and accessories.
Sarvam AI’s voice bots excel in mixed-language conversations and can execute actions such as scheduling appointments and processing payments. The company serves around 50 clients, including Sri Mandir, a popular devotional app with over 10 million downloads. Using Sarvam’s voice technology, the app can guide users through specific rituals at various temples and help them request different blessings. “Try throwing GPT-4 or Claude at Sri Mandir. I can guarantee it won’t work,” said Sarvam co-founder Vivek Raghavan, highlighting the challenges US-based AI models face with Indian languages and regional accents.
While US companies like OpenAI have made strides in voice technology, they have been cautious about releasing it widely, citing concerns about user dependency, potential misuse, and the risk of generating copyrighted content. Nevertheless, India’s AI startups remain optimistic. “AI made for specific use cases, languages, and audiences is more accurate, less expensive to run, and has vastly reduced hallucinations,” said Gnani CEO Ganesh Gopalan, referring to AI’s tendency to generate false information.
Although these startups are primarily focused on the Indian market, they are also exploring international opportunities, including in the Middle East and Japan. Gnani’s voice bots, for example, are already operational in the US, helping a California-based Harley-Davidson leasing company engage with Spanish-speaking customers.
