🏦 Finance AI Buddy

An AI-powered financial analysis assistant with document intelligence, RAG queries, agentic workflows, and data visualization.

---

✨ Features

• 💬 Chat Interface — ChatGPT-like UI with streaming responses, markdown rendering, and code highlighting
• 📂 Document Upload — Upload PDFs, CSVs, Excel, and PowerPoint files for AI analysis
• 🔍 RAG Queries — Ask questions over uploaded documents with retrieval-augmented generation
• 🤖 Multi-Agent Workflows — AutoGen-style agents for financial analysis, document parsing, and visualization
• 📊 Chart Generation — AI-generated matplotlib charts returned as interactive images
• 🪞 Document Mirror — Reconstruct and transform document structures (slides, tables, summaries)
• 🌙 Dark & Light Themes — Glassmorphism UI with smooth transitions
• 💾 Chat History — Persistent conversations with rename/delete support
• ⚙️ Flexible LLM Backend — Works with OpenAI, vLLM, or any OpenAI-compatible API

---

🏗️ Architecture

┌─────────────────────────────────────────────┐
│           Next.js Frontend (:3000)          │
│  app/page.tsx │ app/layout.tsx │ globals.css│
│  TailwindCSS + Framer Motion + React Markdown│
└──────────────────┬──────────────────────────┘
                   │ HTTP / SSE
┌──────────────────▼──────────────────────────┐
│           FastAPI Backend (:8000)            │
│                 api.py                       │
│  ┌─────────────────────────────────────┐    │
│  │ /chat   → Streaming LLM responses   │    │
│  │ /upload → File ingestion + embed    │    │
│  │ /query  → RAG-based Q&A             │    │
│  │ /agent  → Multi-agent workflows     │    │
│  │ /plot   → Chart generation          │    │
│  │ /docs   → Document management       │    │
│  └─────────────────────────────────────┘    │
│  FAISS Vector DB │ File Storage              │
└─────────────────────────────────────────────┘
                   │
          OpenAI-compatible API
    ┌──────────────┼──────────────┐
    │              │              │
 OpenAI API    vLLM Local    Custom API

---

🚀 Quick Start

Prerequisites

• Node.js 18+ and npm
• Python 3.10+
• An OpenAI API key OR a running vLLM server

1. Clone & Install

# Clone the repository
cd Finance-ai

# Install frontend dependencies
npm install

# Install backend dependencies
pip install -r requirements.txt

2. Start the Backend

# Start FastAPI server
uvicorn api:app --host 0.0.0.0 --port 8000 --reload

The API will be available at http://localhost:8000.
API docs: http://localhost:8000/docs

3. Start the Frontend

# In a new terminal
npm run dev

Open http://localhost:3000 in your browser.

4. Configure

Open the Settings panel (⚙️ icon) in the app and enter:

Setting	OpenAI	vLLM Local
API Key	sk-...	any-string (ignored)
API URL	https://api.openai.com/v1	http://localhost:8001/v1
Model	gpt-4	Your model name (e.g. meta-llama/Meta-Llama-3-8B)

---

⚙️ Configuration Examples

OpenAI (Default)

{
  "api_key": "sk-proj-...",
  "api_url": "https://api.openai.com/v1",
  "model": "gpt-4"
}

vLLM Local Server

First, start vLLM:

python -m vllm.entrypoints.openai.api_server \
  --model meta-llama/Meta-Llama-3-8B-Instruct \
  --host 0.0.0.0 \
  --port 8001

Then configure the app:

{
  "api_key": "not-needed",
  "api_url": "http://localhost:8001/v1",
  "model": "meta-llama/Meta-Llama-3-8B-Instruct"
}

Any OpenAI-Compatible API (e.g., Ollama, Together, etc.)

{
  "api_key": "your-api-key",
  "api_url": "https://api.together.xyz/v1",
  "model": "mistralai/Mixtral-8x7B-Instruct-v0.1"
}

---

📁 Project Structure

Finance-ai/
├── app/
│   ├── page.tsx          # Complete frontend (single file)
│   ├── layout.tsx        # Root layout
│   └── globals.css       # Global styles + animations
├── api.py                # Complete backend (single file)
├── package.json          # Frontend dependencies
├── requirements.txt      # Python dependencies
├── next.config.js        # Next.js config with API proxy
├── tailwind.config.ts    # TailwindCSS config
├── postcss.config.js     # PostCSS config
├── tsconfig.json         # TypeScript config
└── README.md             # This file

Runtime directories (auto-created):

Finance-ai/
├── uploads/              # Uploaded documents
├── vector_db/            # FAISS indexes
└── metadata.json         # Document metadata

---

🔌 API Endpoints

Method	Endpoint	Description
POST	/chat	Streaming LLM chat (SSE)
POST	/upload	Upload & ingest documents
POST	/query	RAG-based Q&A over documents
POST	/agent	Multi-agent workflow execution
POST	/plot	Generate charts from data
GET	/documents	List all uploaded documents
DELETE	/documents/{id}	Delete a document

Example: Chat Request

curl -X POST http://localhost:8000/chat \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [{"role": "user", "content": "Explain EBITDA"}],
    "api_key": "sk-...",
    "api_url": "https://api.openai.com/v1",
    "model": "gpt-4"
  }'

Example: Upload Document

curl -X POST http://localhost:8000/upload \
  -F "file=@financial_report.pdf"

Example: RAG Query

curl -X POST http://localhost:8000/query \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What were the Q3 revenue figures?",
    "document_id": "abc-123",
    "api_key": "sk-...",
    "api_url": "https://api.openai.com/v1",
    "model": "gpt-4"
  }'

---

🤖 Multi-Agent System

The /agent endpoint runs an AutoGen-style pipeline:

User Query
    │
    ▼
┌──────────────────┐
│ Document Parser   │  Extracts and structures document content
│ Agent             │
└────────┬─────────┘
         │
         ▼
┌──────────────────┐
│ Data Analyst     │  Analyzes financial data, generates insights
│ Agent             │
└────────┬─────────┘
         │
         ▼
┌──────────────────┐
│ Visualization    │  Creates charts and visual representations
│ Agent             │
└────────┬─────────┘
         │
         ▼
    Final Output (streamed via SSE)

Each agent step is streamed as an SSE event so the frontend can display progress in real-time.

---

🛠️ Tech Stack

Frontend

• Next.js 14 — React framework (App Router)
• TailwindCSS — Utility-first styling
• Framer Motion — Animations
• React Markdown — Markdown rendering
• Lucide React — Icons

Backend

• FastAPI — Async Python web framework
• FAISS — Vector similarity search
• LangChain — LLM orchestration
• Matplotlib — Chart generation
• OpenAI Python SDK — LLM integration

Supported LLM Providers

• OpenAI (GPT-4, GPT-3.5)
• vLLM (any HuggingFace model)
• Any OpenAI-compatible API

---

📄 License

MIT License — feel free to use and modify.