Thu, Sep 18, 2025 | Updated 4:51PM IST

India’s Polyglot AI Is Speaking to a Billion Voices

Times of Bennett | Updated: Sep 05, 2025 17:54
article image
Correspondent Anoushka Saxena

Take a moment outside of your busy schedule, sit in a quiet corner, close your eyes and picture the plight of a grandmother living in a small town in Odisha struggling to figure out the arrival date of her
pension. Suddenly, to her rescue comes a phone. She speaks into it, and a voice replies warmly, in a language that sounds just like her own.

At the same time, a farmer in Bihar snaps a photo of the clouds in his village, saying “Tomorrow, rains?” expecting a hopeful reply. A calm voice replies back in Bhojpuri: “Rain is expected, sow your wheat seeds today.”

Now while all of this sounds like a futuristic dream, it is no longer just a yearning for this has become a reality for many
Indians across the nation. And this dream has been made possible by none other than India’s multilingual AI, which is already reshaping how technology speaks to its people.

For decades, technology in India has spoken to us in a language that often wasn’t our own. English dominated the digital space, leaving millions excluded from the promise of the internet age. But India, home to over 1.4 billion people and over 19,500 mother tongues, was never going to let language become a digital barrier forever. Thanks to BharatGen, Bhashini, KisaanAI and BharatGPT, AI models like these are actively being trained to think, process, and respond in the 22 constitutionally recognized languages of the Eighth Schedule with the fluency of an Indian native speaker by 2026.


Unlike global models that put Indian languages as an afterthought, these homegrown systems put Hindi, Tamil, Odia, Konkani, and other Indian languages right at the centre of this AI revolution. And these models are not mere translation machines. They’re bridging what many call the “cultural gap”, at the hands of which many Indians were ripped of their sense of individuality in the digital space.

The idea of AI models speaking in the language of their user’s choice isn’t just a decision on paper anymore and has already been mobilised into action. BharatGPT, built by CoRover.ai, is the first Indian conversational AI of its kind. You may think of it as an assistant that doesn’t just speak English or Hindi, but dozens of tongues and most importantly,
it works offline and in low-bandwidth settings. That means it runs just as smoothly in a metro city as it does in a rural village with patchy internet.

It’s the voice behind IRCTC’s multilingual ticket booking system, where people can book or cancel tickets without typing a single word. It even powers LIC’s customer support, making policy queries easier to navigate. And if you ever happen to be at the Maha Kumbh 2025, it shows up as “Kumbh Sahayak,” guiding pilgrims in over 11 Indian languages through WhatsApp, apps, and even phone calls.

Meanwhile, out in the fields, KisaanAI has become an indispensable lifeline for farmers. Instead of having to wait long
hours for experts to visit the site and fidget with complicated apps, farmers can now talk to an AI directly in a language they feel most comfortable speaking. This allows them to ask the AI all about crop diseases, mandi rates,
and weather forecasts, even diagnose plant diseases with a simple photo of a leaf.

By mid-2025, KisaanAI had already reached five million farmers across 17 states, proving that technology doesn’t have to be urban, elite, or English-speaking to be deemed powerful. Here, AI can even speak the dialects of the soil.

The government has also joined this nationwide AI revolution with its ambitious Bhashini project under the National
Language Translation Mission
. Under this project lie real-time translation and speech tools that make governance, education, health, and even the courts accessible in multiple languages. Recently, Bhashini partnered with the
railways, ensuring passengers could interact with services in their own tongues. In the judiciary, tools like SUVAS are translating court documents, helping bridge a gap that has long slowed justice. With each rollout, AI is being woven into daily Indian life not as a foreign voice, but as a familiar companion.


For decades, global AI belonged to English, Mandarin, and a handful of European languages. India’s linguistic richness was often treated as a last priority. But now the time has come for Indian startups, researchers, and government missions to show what the country is capable of. With open-source efforts like AI4Bharat, backed by Nandan Nilekani, and
scalable platforms like BharatGPT, India is showing the world what it means to make AI truly inclusive.

For years, AI in India was like a polite guest at a party who only spoke English, nodding along while most of the room stayed silent. But now this guest has learned to dance to Bhojpuri beats, debate in Tamil, whisper in Kashmiri, and joke in Marathi, all in the same night. By 2026, it aims to cover every single one of India’s 22 official languages, and perhaps even more.

And well, this is what you call the coming together of an age of technical upgradation backed by cultural rebellion. It’s India telling the world that intelligence, whether human or artificial, doesn’t have just one accent. It has 1.4 billion and they’ve all just found their voice!

(Currently pursuing 3rd year BA-JMC at Times School of Media, Anoushka Saxena’s portfolio spans podcasts, editorial features, and creative media experiments, all with one goal: to make people stop and listen. Moreover, she’s learning to ask better questions, because she believes every great story begins with one.)