Skip to content

sagarjain03/jolly-LLB

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🇮🇳 JOLLY-LLB — Citizen Advocate AI

Zynd Aickathon 2026 — JOLLY-LLB is a high-availability Citizen Advocate AI built on the Zynd Protocol. It simplifies complex Indian government policies, provides deterministic eligibility verification, and executes automated form filling with a trust-first approach.


🎨 Professional Interface

JOLLY-LLB now features a state-of-the-art Professional Dark Theme dashboard built with Streamlit.

  • Architecture Stepper: Real-time visualization of the AI's internal reasoning stages.
  • Dynamic Profile Builder: Sidebar form to manage user demographics for precise eligibility checks.
  • Actionable AI: Direct "Auto-Fill" integration that launches a background automation agent.

🏗️ Project Architecture

User Query → [ Semantic Search (FAISS) ] 
             → [ Deterministic Rule Gate ] 
             → [ RAG Synthesis (Groq Llama 3.3) ] 
             → [ Zynd Protocol Signing ] 
             → Response + Action Panel

Core Features:

  1. RAG-Powered Policy Navigator: Semantic search across 20+ schemes using Gemini Embeddings and FAISS.
  2. Deterministic Rule Engine: Hardcoded Python logic guards against LLM hallucinations for eligibility checks.
  3. Next Best Action (NBA): Automatically redirects users to alternative schemes if they are ineligible.
  4. Intelligent Form Filler: Playwright-based agent that auto-fills registration forms with 0.5s typing delay for visibility.

🚀 Getting Started

1. Prerequisites

  • Python 3.12+ (Recommended)
  • Chrome or Edge browser (for Playwright)

2. Installation & Setup

git clone <repo-url>
cd jolly-LLB
pip install -r requirements.txt
playwright install chromium

3. Environment Configuration

Create a .env file in the root directory:

GOOGLE_API_KEY=your_gemini_api_key
GROQ_API_KEY=your_groq_api_key
ZYND_API_KEY=your_zynd_api_key

4. Running the Project

Step A: Data Ingestion (One-time)

Embed the scheme data into the FAISS vector database:

python ingest.py

Step B: Launch Backend (FastAPI)

The backend coordinates the form-filling agent and the dummy portals:

uvicorn api.server:app --port 8000

Step C: Launch Frontend (Streamlit)

The main interactive dashboard for users:

streamlit run app.py

📂 Repository Structure

File/Dir Purpose
app.py Professional Streamlit UI with Architecture Stepper.
api/server.py FastAPI backend for form-filling sessions & A2A calls.
agents/form_filler.py Playwright automation logic with slow-typing and success-hold.
logic/eligibility_engine.py Deterministic Python rules for all supported policies.
logic/next_best_action.py Logic for intercepting rejections and finding alternatives.
scripts/query_agent.py Core RAG pipeline with Groq reasoning.
ingest.py Data ingestion pipeline using Google Gemini Embeddings.
templates/ Professional dummy government portals for demo/testing.
tests/ Comprehensive test suite (Agent, Eligibility, NBA).

🛠️ Tech Stack

  • LLM: Groq (llama-3.3-70b-versatile)
  • Embeddings: Google Gemini (models/gemini-embedding-001)
  • Vector Store: FAISS
  • Automation: Playwright
  • Identity/Trust: Zynd Protocol
  • Web: FastAPI & Streamlit

📄 License

MIT — Built for Zynd Aickathon 2026. Jai Hind! 🇮🇳

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors