Sarvam AI unveils ‘Chanakya’ vertical for critical, high-security AI needs

Sarvam, the Indian AI startup, is expanding its business division with a new vertical dubbed “Chanakya.” It is said to focus on applied AI services that claim to provide real-world AI solutions to governments, enterprises, or critical sectors where reliability and security are of the utmost priority.

Sarvam shared an X (formerly Twitter) post announcing the launch of Chanakya. The company revealed that it was quietly building full-stack AI capabilities for the past year to solve high-impact problems, such as affecting national interests or large, complex organisations. 

“Applying our full-stack AI to problems of national consequence and complex enterprises. Now we're scaling the vertical behind it - Chanakya,” said the company in a post. It was further highlighted that the AI system will be deployed to air-gapped environments, which are isolated from the internet for security.

Chanakya is designed to analyse multiple types of data, including text and images, and autonomously run AI workflows reliably in real-world, high-stakes scenarios. 

Built with strong security features, the company is targeting organisations that do not rely on public cloud infrastructure and work in highly secure environments. It says that the system supports “dual use,” meaning it can be deployed across both enterprise applications and strategic sectors such as defence and national security.

As it expands its enterprise footprint, the AI startup is rapidly scaling its AI services with advanced offerings. At the Indian AI Impact Summit 2026, Sarvam launched its first AI models, Sarvam-30B and Sarvam-105B, with support for 22 Indian languages. The company gained much traction for their performance in comparison to global AI models.

It also showcased AI-powered smartglasses, dubbed Sarvam Kaze, and launched its AI-first app called “Indus”, and much more. Sarvam has also made a global push, joining Nvidia’s Nemotron Coalition alongside leading AI firms to build next-generation open foundation models, marking its entry into a broader international effort to shape the future of open and sovereign AI.