
Sarvam AI is set to launch an innovative voice-enabled artificial intelligence app specifically designed for India's multilingual demographic. Co-founder and CEO Pratyush Kumar emphasized in a recent interview that the key to enhancing India's position in the global AI landscape lies in consumer acceptance rather than merely developing advanced models. The forthcoming application will enable users to communicate with a locally trained large language model through both text and voice in various Indian languages. Kumar stated, "Once the app is launched, users will be able to engage in conversations in English and Indian languages." He highlighted that this app is intended to serve as a mass-market alternative to existing Western AI assistants, offering general-purpose AI capabilities right at users' fingertips. As India's aspirations in AI grow, the focus is shifting towards translating sovereign models into practical applications. While local companies have made strides in training foundational models, widespread adoption remains a challenge. Sarvam aims to bridge this gap by introducing a consumer-centric platform that aligns with local usage patterns. Kumar noted, "India is a voice-first nation," asserting that voice interaction is essential given the country's linguistic diversity and varying literacy levels. The app is designed to address everyday inquiries, ranging from educational support to agricultural advice and local searches. Kumar explained that the app will allow a student to ask questions about their studies or a farmer to inquire about government schemes, showcasing a practical approach to AI rather than just a novel concept. Sarvam's initiative signifies a crucial transition in India's AI sector, moving from experimental phases to actual deployment, as both startups and policymakers prioritize accessibility. The company has developed its model utilizing domestic computational resources, with government assistance, in a bid to lessen dependence on international platforms. According to Kumar, "India should be building its own solutions," as the nation enters a "post-AI world" where large language models become integral to daily life. On February 18, Sarvam launched two large language models, Sarvam-30B and Sarvam-105B, during the India AI Impact Summit. Both models are set to be released as open source, aiming to encourage adoption among developers, businesses, and government entities looking for alternatives to foreign AI systems. Since early February, the Bengaluru-based startup has rolled out various AI tools for tasks such as speech transcription, dubbing, and translation across 22 Indian languages, as well as optical character recognition and on-device AI solutions. Kumar cautioned that while developing the model is a significant milestone, the greater challenge lies in widespread implementation. He stated, "The transition from ‘I trained a model’ to ‘I’m using the model at population scale’ is a problem worth solving," emphasizing the need for practical application in a nation of over 1.4 billion people.
Fizz, a social app that allows users to post anonymously, has made its international debut in Saudi Arabia, marking a si...
TechCrunch | Apr 03, 2026, 22:50
Anthropic has officially announced the establishment of a new political action committee (PAC), signaling its commitment...
TechCrunch | Apr 03, 2026, 21:00
In a bold move, Elon Musk is requiring banks and consulting firms involved in SpaceX’s upcoming initial public offering ...
Ars Technica | Apr 03, 2026, 21:20The expansion of data centers has sparked significant public debate, with a recent poll from Harvard and MIT highlightin...
TechCrunch | Apr 03, 2026, 19:40
Donald Trump is encountering major setbacks in his quest to rapidly expand AI data centers across the United States, a k...
Ars Technica | Apr 03, 2026, 20:50