How AI is Powering the Future of Citizen Services

India’s diversity makes it a vibrant and culturally rich nation, strengthening it from within. The tapestry of languages, religious diversity, and geographical and regional symbiosis adds to pluralism within the country and on a global stage.

Despite this diversity, the development of technology and its adoption in the country, however, has been largely anchored in the English language.

As artificial intelligence gets increasingly embedded in services including healthcare, education, governance, and other essential services, the need to diversify and make these technologies respond to Indic languages and dialects becomes all the more imperative.


However, this has already started changing and at a significant scale. India’s AI story is ushering in a new direction, and it has the potential to redefine the country’s AI advantage.

This is what was on the agenda in this third episode of Blueprint to Bharat, where senior journalist Sonia Singh, from The Economic Times, had an immersive conversation with Ankush Sabharwal, founder and CEO of CoRover, and Harsh Dhand, AI partnerships lead for APAC at Google India.


“He just taps a button on his phone and speaks naturally in his native language. In fact, he even mixes in local phrases like he's talking to a friend. The system instantly understands his intent,” Singh’s hypothetical scenario was about a father trying to book train tickets amid a festive season, illustrating the potential for inclusive AI adoption in the country.

And this is precisely what Sabharwal’s CoRover.AI has been enabling!

Building AI to be inclusive:

Take CoRover for example. The conversational AI platform has powered three trillion interactions for over a billion users across a hundred-plus Indic languages.

The company, which uses Google Gemini’s AI models, is taking the challenge of AI inclusion beyond English-speakers head-on, and has had a great run. How did it come into being, though?

Sabharwal recounted a story from when he was travelling to Bengaluru on a train, a 36-hour journey to his first job, where he witnessed a young boy seeking help for his mother who had apparently suffered a cardiac arrest.

“The boy could not speak Hindi or English, while the co-passengers tried to help. The language barrier and the challenge of not being able to use technology to seek help despite the government’s health infrastructure being available were evident,” Sabharwal said, adding that he was getting goosebumps recounting the story. The incident stayed with him, and being a doer, CoRover has been his “bit” to address the challenges the boy and his mother faced that night.

“I had thought then that we would have to remove the barriers of language and of technology,” he added. “It is important because only a small percentage of Indians understand English, and the technology designed to cater to this percentage leaves out a significant number of Indians.”

Supporting the foundation: Indic language dataset

It is precisely to strengthen this effort of AI inclusion that Google has collaborated with the Indian Institute of Science Bangalore on initiatives like Project Vaani, which has collected speech sets of over 100 Indian languages so far.

“We, at Google, realised two things early on. One, the immense potential AI has in terms of problem solving, and second, that information is available, accessible, and consumed in all languages,” Dhand said.

He said that putting the two things together, the focus naturally is inclined towards ensuring that technology, used by millions in India to improve their quality of life, is made available in all languages.

“The language data sets we have collected through Project Vaani are open-sourced and available to the entire Indian ecosystem,” he said. “We are also providing a lot of research grants in this domain, like the one of $2 million we provided to the Indic Language Research Technology Centre at IIT Bombay,”

We are really looking forward to a future where technology and language are seamless, and it integrates into the lives of people every day, Dhand said.

Why is it important?

One of the most important aspects and implementations of multilingual AI is making public services conversational. Or in simpler terms, we can say it is about making governance accessible to an average Indian citizen, who may not have the means or the knowledge to browse government websites or fill complex forms therein.

The voice technology of CoRover has bridged many barriers. It can support solutions that enable people to navigate complex systems easily by speaking in their own native language - whether they want to register a grievance or find a government policy to make a livelihood.

“If they get the information, using our voice technology, at the right time, get the policy at the right time, they will upskill themselves, they will get educated. They will do good for themselves, for their family and the society,” Sabharwal said.

Incorporating diversity at scale:

When asked about what building multilingual AI models in Indic languages means

Dhand said that language was not just about words, and often carries cultural, historical, and geographical context. That awareness remains, he said, at the foundation of building multilingual AI models in Indic languages.

“We understand that for AI models to work seamlessly in India, they must understand dialects, code-mixed conversations, regional expressions, and local context,” he said. “With all these things in mind, we look at language inclusivity in our model development and ensure we understand the present-day use case of that particular language, including all its dialects and cultural nuances.”

Beyond this, Dhand explained, Google keeps incorporating prompts in Indic languages, specific to regions. Taking the efforts further, Google is also in collaboration with some of the top universities in Africa and India for a project called Amplify.

“The idea is exactly the same: how do we bridge the gap between the current-day AI models and include the language nuances that are there,” he said.

The next frontier:

Development of such models is crucial in the sense that they can not only help make lives better, but can also bridge the gap between the citizen and the government. They empower the citizens to interact with the government for something that is their right.

“Can CoRover, and other organisations in partnership with Google, actually help break down the barrier of a citizen’s rights and the rights that are available to every Indian?” Singh asked.

The thought of it keeps Sabharwal “restless” as well, and he thinks very little has been done while a lot more needs to be executed.

“But with a mighty partner like Google, the onus is on us to deliver and solve this for India,” he said, “But the good part is, we think big and we ensure that it reaches the bottom of the pyramid.”