NagrikAI

Its an agentic AI copilot that empowers every Indian citizen to access government schemes and services seamlessly, in their own language, on any device.

Comment

Inspiration

Despite India having 700+ government welfare schemes, millions of citizens especially in rural areas remain excluded due to language barriers, fragmented portals, and complex forms.
We were inspired by the idea of turning AI into a copilot that doesn’t just answer queries but actively completes bureaucratic tasks for citizens, making governance truly inclusive.

What it does

NagrikAI is a multi-agent AI system that acts as a personalized copilot for citizens. It:

  • Verifies eligibility for welfare schemes.
  • Auto-fills forms using DigiLocker and Aadhaar.
  • Routes grievances and tracks their resolution.
  • Provides contextual, multilingual information.
  • Sends reminders and status updates.

All of this happens in 22+ Indian languages, over low-bandwidth channels like WhatsApp, SMS, and IVR ensuring accessibility for digitally underserved populations.

How we will build it

We will design NagrikAI with a modular multi-agent architecture powered by:

  • IBM Agent Development Kit (ADK) for orchestration.
  • Granite 34B + finetuned Llama3 for multilingual understanding.
  • Bhashini + IndicTrans2 for translation.
  • Weaviate + custom RAG pipeline for retrieval.
  • LayoutLMv3 OCR for document understanding.
  • WhatsApp Business API + IVR for delivery channels.

All workflows will be privacy-preserving and compliant with India’s DPDP Act.

Challenges we might face

  • Handling 22+ languages with code-mixed queries is extremely complex.
  • Indian government documents have highly varied layouts, requiring OCR fine-tuning.
  • Ensuring scalability on low-bandwidth rural networks will be a major technical hurdle.
  • Balancing automation with citizen trust and transparency requires careful design.

Excited for

  • Exploring how agentic architectures can go beyond chatbots to handle end-to-end citizen services.
  • Prototyping solutions that could reach 600M+ underserved citizens, including those with only feature phones.
  • Experimenting with integrations into India’s Digital Public Infrastructure (UMANG, Bhashini, AgriStack).

What's next for NagrikAI

  • Disaster response module integrated with NDMA for emergency relief.
  • Healthcare navigation with ABHA for appointment booking and telemedicine.
  • Financial inclusion workflows for PMJDY and simplified banking access.
  • Long-term vision: voice-first governance, predictive eligibility, and blockchain-based verification for trustless service delivery.

Built With

Share this project:

Updates