Voice-Enabled Bots: How This AI Startup is Scaling AI Adoption in India’s Multilingual Market
Ibrahim Awotunde
In a country like India, where there are 22 official languages and over 19,000 dialects, the traditional text-only AI chatbots face significant limitations. Recognizing this challenge, Bengaluru-based startup Sarvam has introduced a groundbreaking solution: voice-enabled AI bots that support more than 10 Indian languages.
This move aims to revolutionize AI adoption in India by allowing users to interact with AI in their native tongues, a critical factor in a linguistically diverse nation.
Breaking Language Barriers with AI
Sarvam’s co-founder, Vivek Raghavan, highlights the difficulty of typing in Indian languages, which often deters people from engaging with text-based AI models. “People prefer to speak in their own language. It’s extremely challenging to type in Indian languages today,” Raghavan noted. To address this, Sarvam is rolling out AI voice-enabled bots designed to enhance customer support across various industries.
One of the early adopters of Sarvam’s technology is Sri Mandir, a startup offering religious content. Using Sarvam’s AI agent, Sri Mandir has successfully processed over 270,000 transactions, demonstrating the potential of these bots in real-world applications.
The versatility of Sarvam’s voice-enabled AI agents extends beyond apps and can also be integrated into WhatsApp and traditional voice calls, making them accessible to a broader audience.
Affordable and Scalable AI Solutions
Sarvam’s AI agents are built on a robust small language model called Sarvam 2B, which has been trained on an extensive dataset of 4 trillion tokens. Despite concerns around the use of synthetic data—which can sometimes introduce inaccuracies—Sarvam has taken measures to refine and improve the quality of this data before using it for training. This approach has enabled the startup to offer its AI solutions at a fraction of the cost of industry standards. The AI agents are priced competitively, starting at just ₹1 (approximately 1 cent) per minute of usage, making them accessible to businesses of all sizes.
In addition to its voice-enabled bots, Sarvam is also introducing an audio-language model named Shuka, which is built on its proprietary Saaras v1 audio decoder and Meta’s Llama3-8B Instruct. This model is open-sourced, allowing developers to create custom voice interfaces using Sarvam’s translation and text-to-speech (TTS) modules.
Empowering Legal Professionals with AI
Sarvam is not stopping at voice-enabled bots. The startup is also launching “A1,” a generative AI workbench specifically designed for legal professionals. This tool can perform a variety of tasks, including looking up regulations, drafting documents, redacting sensitive information, and extracting critical data, all of which streamline the legal workflow and enhance efficiency.
Contributing to India’s AI Ambitions
Sarvam’s initiatives align with the broader goals of India’s “IndiaAI” program, which aims to establish a sovereign AI infrastructure tailored to the country’s unique needs. As global interest in sovereign AI grows, with significant investments from the United States and China, India is also stepping up its efforts.
The IndiaAI program includes plans for a supercomputer powered by at least 10,000 GPUs and the development of language-specific models like Bhashini, which aims to democratize access to digital services across various Indian languages.
Raghavan expressed Sarvam’s readiness to contribute to these national efforts, stating, “If the opportunity arises, we will work with the government.” With its innovative solutions and commitment to advancing AI in India, Sarvam is positioning itself as a key player in the country’s AI landscape.
By addressing the linguistic diversity and technological needs of the Indian market, Sarvam is not only making AI more accessible but also setting a new standard for how technology can be tailored to meet the specific needs of a diverse population.
