Inspiration

In India, thousands of citizens remain unaware of the benefits they are eligible for because government schemes are complex, scattered, and mostly in English/Hindi. This inspired us to build Sahayak AI – an AI assistant that explains schemes in simple, local languages, so every citizen can access their rights without barriers.

What it does

Sahayak AI is a Gen-AI powered assistant that makes government schemes simple, clear, and accessible for every citizen.

  1. Explains schemes in easy-to-understand language.

  2. Provides eligibility checks (e.g., “Am I eligible for PM Kisan?”).

  3. Gives step-by-step guidance on how to apply.

  4. Supports multiple Indian languages for wider reach.

Offers voice-enabled responses for citizens with low literacy.

How we built it

  1. Collected government scheme FAQs and documents.

  2. Preprocessed data into Q&A format using Kaggle.

  3. Designed a Retrieval-Augmented Generation (RAG) pipeline for accurate responses.

  4. Built conversational flows with IBM Agent Development Kit (ADK).

  5. Added multilingual translation using IBM Watson APIs.

  6. Enabled voice support for low-literacy users.

    Challenges we ran into

  7. Collecting structured scheme data from scattered sources.

  8. Handling multiple languages with accuracy.

  9. Optimizing the assistant for fast, relevant answers

    Accomplishments that we're proud of

  10. Built a working AI assistant prototype that can explain government schemes in multiple Indian languages.

  11. Successfully integrated IBM Agent Development Kit (ADK) with a RAG pipeline for accurate and reliable answers.

  12. Created a simple, user-friendly experience that can support both text and voice queries.

  13. Managed to convert complex policy documents into citizen-friendly explanations.

  14. Demonstrated how Gen-AI can bridge the language and digital divide for millions of people.

  15. Showed that even with limited time and resources, it’s possible to create a project with real social impact.

    What we learned

  16. How to integrate Gen-AI with real-world datasets.

  17. Building multilingual conversational AI is challenging but highly impactful.

  18. Learned about the power of RAG for government/official document retrieval.

    What's next for Sahayak AI (Citizen Scheme Assistant)

  19. Expand Coverage → Add support for all central and state government schemes, starting with the most widely used programs.

  20. More Languages & Dialects → Scale to cover 22+ Indian languages and popular dialects, making the assistant accessible to every citizen.

  21. Seamless Access → Deploy as a WhatsApp/Telegram bot so users in rural areas can access it easily without downloading new apps.

  22. API Integration → Connect with platforms like UMANG, DigiLocker, and CSC portals to directly help citizens apply for schemes.

  23. Smart Document Support → Enable users to upload documents (like Aadhaar, land records) and check if they meet eligibility requirements.

  24. AI-Driven Personalization → Provide scheme recommendations based on user profile, region, or occupation (e.g., farmer, student, senior citizen).

  25. Accessibility Features → Strengthen voice-first interaction for visually impaired and low-literacy users.

Built With

Share this project:

Updates

Submission history