Bhashini: India made language translation a public rail any app can plug into

India's digital boom ran in English and Hindi while hundreds of millions of its users transact, complain and buy in Tamil, Bangla, Marathi and a dozen other languages. In 2022 the government launched Bhashini at Gandhinagar, re-christening the National Language Translation Mission into a platform that treats translation the way UPI treated payments: shared public rails, open to anyone (Economic Times).

The data problem came first — Indian-language training data barely existed. Bhashini's answer was BhashaDaan, a crowdsourcing front door with four drives: Suno India (transcribe what you hear), Likho India (write), Bolo India (speak), Dekho India (validate). Contributors earn Bhasha Samarthak recognition badges, and the output is an open-source language database any developer can build on (Times of India).

The proof moment came in December 2023, when the Prime Minister's Hindi speech at Kashi Tamil Sangamam in Varanasi carried a real-time Tamil translation powered by Bhashini (The Hindu). A year later ONDC co-developed Saarthi, a reference buyer-side app for e-commerce that launched in five languages with plans to scale to all 22 languages Bhashini provides (The Hindu), while MeitY ran bids for universities to sharpen translation accuracy in domains like health and education (Economic Times).

Language became a public rail — every app just plugs in

Crowdsourcing turned the missing Indian-language training data into a mass participation product — recognition badges instead of wages.

Open APIs shifted the build cost: a bank, a marketplace or a state portal gets multilingual features almost free instead of each building its own translation stack.

The state stayed at the rail layer and left the app layer to everyone else, which is what made UPI cheap to adopt and hard to compete with.

A Prime Minister's speech translated live into Tamil was a public demonstration that the rail works at speaking pace, not demo pace.

When one constraint blocks every player, solve it once as shared infrastructure — build the layer under everyone's products, the way UPI did for payments.

By mid-2024 MeitY was soliciting university bids to raise accuracy and fluency for specific language pairs and critical sectors (Economic Times). In September 2024 ONDC shipped Saarthi, co-developed with Bhashini, as a reference app meant to seed many multilingual buyer apps (The Hindu). The pattern — state builds the rail, market builds on it — is spreading from payments and identity to language itself.