India has picked one vendor to build the conversational front door to its government services. Bengaluru-based agentic AI platform CoRover said on Wednesday that it has won a tender from the National e-Governance Division (NeGD) under the Ministry of Electronics and Information Technology to develop a unified AI assistant and bot platform, with implementation set to begin on DigiLocker, the country's digital document wallet.
The tender drew more than 80 participants, including global consultancies Deloitte, EY, PwC and KPMG, IT services firms Tata Consultancy Services, NEC and Kyndryl, and AI companies Jio Haptik, Fractal and Gnani. CoRover said it secured the highest technical score and emerged as the only L1 bidder, making it the sole selected partner; a summary of the evaluation puts its quality-and-cost score at 100 against Kyndryl's 91.24. The project includes a four-month implementation period and 12 months of maintenance support.
Founder and CEO Ankush Sabharwal told The Economic Times that implementation would commence with DigiLocker, potentially within the next four months. DigiLocker is where Indian citizens store digitally verified versions of driving licences, vehicle registration certificates, academic marksheets and other documents, and it is the kind of high-volume, authentication-heavy service where a conversational layer either earns trust quickly or fails publicly.
CoRover is not an unknown quantity in Indian government tech. The company built AskDISHA, the AI chatbot deployed by Indian Railway Catering and Tourism Corporation in October 2018 - the first AI assistant fielded by an Indian public sector enterprise. By the company's own account, AskDISHA has handled more than 10 billion interactions for roughly 178 million passengers, processing about 150,000 queries a day at around 90% accuracy. Those are company-reported figures, not independently audited.
The company's broader platform pitch rests on BharatGPT, its sovereign large language model built for Indian languages and contexts, alongside voice bots, video bots and virtual assistants spanning more than 100 languages and channels from websites to messaging apps. CoRover also allows clients to run its systems inside controlled environments on their own data - an increasingly decisive requirement for government deployments where data residency is non-negotiable.
The deal matters for two reasons. First, it moves government AI in India from scattered department-level bots toward a shared platform, which is how conversational AI reaches hundreds of millions of citizens rather than early adopters. Second, it is a signal of how governments are starting to buy AI at all: through quality-and-cost tenders against global integrators, with sovereignty and controlled deployment weighted heavily.
What the announcement did not disclose is the contract value, and the accuracy and cost-reduction claims around CoRover deployments remain company figures. Large-scale government assistants also face hard, unglamorous problems - dialect and code-mixed language, hallucinated answers to entitlement questions, and the reputational cost of a wrong answer about a citizen's documents.
Those caveats aside, the direction is set. If the DigiLocker rollout lands within four months as planned and holds up under real citizen traffic, expect the same platform to spread across other Indian government services quickly - and expect other governments watching how a national-scale public AI assistant is procured and run.
Comments (0)
Log in to join the discussion
Log InNo comments yet