Hyderabad: Government services are going to be more accessible to citizens with Digital India BHASHINI Division (DIBD) and NITI Aayog joining hands to expand the use of multilingual, voice-based Artificial Intelligence in governance.
Amitabh Nag, CEO of the DIBD, signed a Statement of Intent (SoI) with Shoyabahmed Kalal, Director, NITI Aayog, on August 12.
What is the initiative?
BHASHINI for Seva/Sanchalan initiative is aimed at integrating AI-powered language technologies into NITI Aayog’s digital platforms and public service workflows.
The initiative could make AI-driven government services less dependent on English-language interfaces and bring voice and multilingual capabilities closer to citizens at the last mile.
Breaking language barriers
For India, where language remains one of the biggest barriers to digital access, the experiment is essentially about making AI speak the citizen’s language, enabling them to avail government services without a language barrier.
At the heart of the collaboration is BHASHINI’s ability to translate, transcribe and process information across multiple Indian languages. This will allow citizens to access government information and services in their preferred languages, while also enabling government departments to communicate with people in those languages.
DIBD powering 800 govt websites
DIBD was launched by Prime Minister Narendra Modi during Digital India Week in Gandhinagar in July 2022. DIBD is the National Language Translation Mission. Its objective is to enable every citizen to access digital public services and the internet in their preferred Indian language, including through voice-based interaction.
The platform currently powers more than 800 government websites and has recorded over 9 billion cumulative AI inferences, with more than 24 million AI inferences processed every day, according to the government.
BHASHINI supports 36 Indian text languages, 23 Indian voice languages and 35 international languages.
Recently, the Delhi Legislative Assembly has also invited the DIBD to conduct a pilot trial of its AI-powered live transcription and translation system during its proceedings.
From translation to voice-based AI
The NITI Aayog partnership is a step towards using this existing language AI infrastructure for public administration. The partnership will bring together several components of the BHASHINI ecosystem, including Udyat, Mitra, AppMitra, Sahayogi and Pravakta.
These tools will be used to integrate language translation APIs into NITI Aayog’s digital platforms, document management systems and communication workflows. The organisations will also work on improving translation models and developing applications for sectors that require specialised terminology.
Developing voice bots
One of the more AI-focused components is the development of reference applications and voice bots for specific NITI Aayog sectors.
This could allow government services to move beyond text-based interfaces and towards voice-first interactions, potentially making them easier to use for citizens with limited digital literacy.
The collaboration will also use BhashaDaan, through BHASHINI Sahayogi, to collect and curate linguistic data. Officials and citizens will be encouraged to contribute text, audio and video samples that can help improve India’s language AI models.
AI at the grassroots
The partnership also builds on NITI Aayog's NITI TARA (Toolkit for Analysis, Review & Action) platform, which is being used to turn data into actionable insights for district and block-level governance.
According to the government, more than 200 capacity-building sessions have been conducted through NITI TARA, reaching over 4,000 officials.
Another component is SMRITI (System for Meeting Recording, Information & Transcription Interface), developed in collaboration with BHASHINI.
SMRITI uses AI to automate meeting minutes and multilingual translation. A government meeting can be recorded, transcribed and translated without officials having to manually prepare the entire documentation.
The goal is to create a government ecosystem where AI is not limited to chatbots or translation tools but becomes part of everyday administrative workflows.
Simplifying technical information
The larger challenge, however, is not simply translating government content.
Government communication often involves technical, legal and sector-specific terminology, where a direct word-for-word translation may not be enough. The focus on specialised terminology, voice interfaces and the continuous collection of linguistic data is an attempt to address this issue.
