BHASHINI is India's national language technology mission, formally the National Language Translation Mission. The Prime Minister launched it on 4 July 2022 during Digital India Week at Gandhinagar, Gujarat. Its purpose is to make digital content and public services usable in Indian languages rather than only in English, and it does this by offering a shared stack of language services — machine translation, speech recognition, speech synthesis and optical character recognition — through open application programming interfaces that any department, start-up or developer can plug into. Its architecture is deliberately voice-first: the design assumption is that a large share of first-time internet users in India will speak to a service rather than type into it. The mission covers the 22 languages listed in the Eighth Schedule of the Constitution. Data for the underlying models is gathered partly through BhashaDaan, a crowdsourcing platform on which volunteers contribute text, audio and video in their own languages, since Indian languages are severely under-represented in the datasets on which global AI models are trained.

The barrier to digital public services in India is often linguistic rather than technological. A scheme portal, a grievance form or an advisory that exists only in English excludes the citizen it is meant to reach. By routing NITI Aayog's platforms and communication workflows through BHASHINI's APIs, the pact tests whether language access can be treated as shared public digital infrastructure — built once at the centre and reused by every department — rather than as a translation cost each ministry absorbs separately. This is the same design logic that governed Aadhaar, UPI and DigiLocker, which is why BHASHINI is usually discussed as a layer of India Stack. For NITI Aayog the immediate gain is operational: SMRITI transcribes and translates meeting records, cutting the manual effort in an institution whose principal output is documentation. The wider significance is that a policy body, not just a service-delivery ministry, is now the testing ground for multilingual AI in government.