From the lab to the front lines of work
Foundation models built for Indian languages — Hindi, Odia, Bengali, Tamil, Telugu, Marathi and beyond — with sovereignty and frugality at the core.
Ask questions of your own documents and data in natural language, across Indian languages — with sourced, cited answers.
Production-grade agents for real workflows — document processing, support, internal knowledge — deployed securely on your cloud.
हिन्दी
ଓଡ଼ିଆ
বাংলা
தமிழ்
A billion people's languages, treated as first-class.
The name itself — Tanu (body, form) and Manasa (mind) — comes from Indian thought: giving form to intelligence, and mind to machines.
That philosophy shapes the work: models built around Indian languages, idioms and knowledge systems rather than translated toward them — from the 22 scheduled languages to the tribal and low-resource ones AI has ignored.
Our story →One lab. Three disciplines.
Foundation models, agentic systems, multimodal and scientific AI, and AI safety — published as we learn. Frugal, efficient development is our discipline.
Read our research →Models trained, aligned and governed in India — made achievable by cost-efficient development. The goal: an India that builds its own AI rather than importing it.
The vision →As an AWS Partner we take enterprises from pilot to production — strategy, agents, and cloud infrastructure that holds up in the real world.
Enterprise AI →
"India should not only use the world's AI — it should build its own, in its own languages, on its own terms."
Building Bharat's next foundation models — from Odisha, for India, for the world.