Inspiration

In India, millions know how to earn money, but not how to manage it. We noticed that financial literacy resources online are text-heavy, English-centric, and inaccessible for rural users who prefer voice-based learning. We wanted to create a tool that teaches finance in local languages through audio, visuals, and simple conversations like having a friendly guide in your pocket. Financio was born to make finance easy, human, and reachable for everyone.

What it does

Financio is a voice-first financial literacy app with micro-lessons, quizzes, and an AI assistant Mitra that answers finance questions in local languages. Users can learn UPI safety, banking basics, saving habits, government schemes, loans, insurance and more using audio lessons + picture cards, even offline. It personalizes lessons by profession and uses agentic AI to guide, nudge, and remind users like a coach.

How we built it

Frontend in Flutter for lightweight Android performance Firebase for backend (Firestore, Storage, Authentication, Analytics) Lesson packs stored as ZIP (audio + images + metadata) for offline mode Node.js Cloud Functions for chatbot routing + RAG Google STT/TTS for voice input/output Agent layer with Learning Coach, Finance Assistant & Reminder Agent Local DB using sqflite to store progress, quiz scores, downloads

Challenges we ran into

Handling accurate speech-to-text for regional accents Designing UI usable without heavy reading dependency Building offline-first content delivery Reducing hallucination in AI responses → solved with FAQ + RAG Keeping audio + image lesson packs lightweight for low-end devices

Accomplishments that we're proud of

1.Built a multilingual, voice-based learning experience 2.Made finance education accessible without heavy text literacy 3.Integrated online+offline learning flow successfully 4.Designed agent behavior to guide users beyond static chatbot answers 5.Created a working prototype in limited time

What we learned

Simple UX matters more than features for first-time digital users Agentic AI creates real impact when it acts, not just answers Localization and trust-building design choices are crucial Offline-first architecture is challenging but essential for Bharat

What's next for Financio

Add more regional languages & lesson packs Expand agent capabilities to complete workflows like form filling Build scam-simulation training for fraud awareness Launch community pilot with 50–200 users Release a dashboard for NGOs/government programs Integrate WhatsApp Lite version for wider adoption

Built With

  • cloud-functions
  • dart
  • firebase
  • flutter
  • google-stt/tts
  • node.js
  • openai/llama-api
  • rag
  • sqflite
  • supabase
  • tensorflow-lite
  • vector
Share this project:

Updates