AutoAnalyst AI 📊

AutoAnalyst is an autonomous Business Intelligence agent capable of ingesting raw datasets (CSV, Excel), performing Python-based statistical analysis in a sandboxed environment, and generating boardroom-ready reports with strategic insights and visualizations.

It features a secure SaaS architecture with Google Authentication, user history persistence, and programmatic PDF export capabilities.

🚀 Key Features

• Generative BI: Transforms raw data into "McKinsey-style" executive summaries, KPI tables, and deep-dive analysis.
• Autonomous Coding: Uses a LangChain agent running in a secure E2B Sandbox to write and execute Python code for data visualization (Line, Bar, Donut charts).
• Multi-File Logic: Intelligently handles multiple datasets, detecting overlapping dates or analyzing distinct files separately.
• Professional UI: Glassmorphism-inspired React dashboard with Markdown rendering, interactive charts, and drag-and-drop file staging.
• Secure Infrastructure:
  • Auth: Google OAuth2 + JWT Session management.
  • Database: MongoDB (Motor/AsyncIO) for storing chat history and analysis results per user.
  • Export: Client-side programmatic PDF generation (jsPDF) for high-fidelity reports.

🛠 Tech Stack

Frontend

• Framework: React (Vite)
• Styling: Tailwind CSS + Lucide Icons
• State: Axios + React Hooks
• Visualization: Custom Markdown Components (remark-gfm) + Base64 Image Rendering

Backend (The Brain)

• API: FastAPI (Python)
• AI Orchestration: LangChain (Graph Workflow)
• Code Execution: E2B Code Interpreter (Sandboxed Python Environment)
• Database: MongoDB Atlas
• Security: OAuth2, BCrypt, JWT

📸 Screenshots

🏗️ Architecture

1. User Upload: Files are sent to the FastAPI backend.
2. Agent Reasoning: The LLM (Gemini/OpenAI) plans a "Visualization Strategy" (e.g., Top 5 Rule).
3. Sandboxed Execution: Python code is generated and executed inside an E2B Sandbox to ensure security and prevent hallucinations.
4. Response Construction: The agent returns a structured Markdown report + Base64 encoded images.
5. Rendering: The React frontend parses the Markdown into UI cards and renders the charts.

⚡️ Getting Started

Prerequisites

• Node.js & npm
• Python 3.10+
• MongoDB Atlas URI
• Google Cloud Console Credentials

1. Backend Setup

cd backend
python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows
pip install -r requirements.txt

# Create a .env file with:
# MONGO_URL=...
# GOOGLE_CLIENT_ID=...
# SECRET_KEY=...
# E2B_API_KEY=...
# GOOGLE_API_KEY=...

python main.py