Enterprise voice AI firm Gnani.ai has announced the launch of Prisma v2.5, its advanced speech-to-text model. Designed specifically for the Indian market, the model supports twelve languages and is trained on an extensive dataset of 14 million hours of proprietary Indic speech. This release aims to solve complex transcription challenges in critical sectors by offering superior accuracy and understanding of local linguistic nuances.
A Model for India's Linguistic Reality
Prisma v2.5 is engineered to handle the realities of Indian conversations, including dialect variations, ambient noise, and natural code-switching between languages. According to CEO Ganesh Gopalan, unlike models built for ideal studio conditions, Prisma is optimized for calls over compressed networks and diverse accents. This focus on real-world application sets it apart from many existing automatic speech recognition systems.
The company reports that internal and third-party benchmarks show Prisma outperforming both global and local competitors like Microsoft and ElevenLabs. Hosted on local Indian data centers, the model also provides significantly lower latency, making it ideal for real-time voice applications. One early retail client has already migrated from a global provider, citing these performance benefits.
Strategic Growth and Market Expansion
This product launch follows Gnani.ai's recent successful Series B funding round in March, where it raised $10 million. The investment, led by Aavishkaar Capital with participation from existing backer InfoEdge Ventures, is fueling the company's strategic growth. These funds are being allocated towards research and development, talent acquisition, and expanding its customer base into new verticals.
While the current version of Prisma focuses on Indian languages, Gnani.ai has clear global ambitions for the technology. The company plans to extend its reach by launching the model in international markets, including Japan, the Philippines, and the Middle East. This expansion strategy aims to position Prisma as a leading global solution with deep localization capabilities.
Competing in the Sovereign AI Arena
Gnani.ai is a key player in India's burgeoning sovereign AI ecosystem, competing with other homegrown firms like Sarvam AI and Fractal Analytics. The startup is also backed by the government's IndiaAI Mission, which aims to foster the development of AI models tailored for India. This competitive landscape highlights a national push towards technological self-reliance and innovation in artificial intelligence.
The case for sovereign AI has been strengthened by recent geopolitical events, such as US export restrictions on advanced AI models from companies like Anthropic. Such developments expose the risks of depending on foreign-controlled technologies and underscore the importance of developing secure, locally-controlled AI systems. Gopalan noted that these events validate the need for countries to build their own AI models with robust safeguards.
The launch of Prisma v2.5 marks a significant milestone for Gnani.ai, reinforcing its commitment to delivering highly accurate and context-aware voice AI for the enterprise market. By addressing specific linguistic challenges and leveraging strategic funding, the company is well-positioned for both domestic and international growth. This initiative not only enhances Gnani.ai's competitive edge but also contributes to India's broader goal of achieving AI sovereignty.