Enterprise Knowledge Base Assistant

A production-ready intelligent knowledge management system using LlamaIndex that enables organizations to query, analyze, and extract insights from multiple data sources (documents, databases, APIs, web content) through natural language.

Key Features

• Multi-Source Data Ingestion: Documents (PDF, Word, Markdown), databases, APIs, web content
• Advanced Query Engines: Sub-question decomposition, SQL generation, router query engine, hybrid search
• Hybrid Search: Vector similarity + keyword search for better retrieval
• Query Analytics: Track query patterns, popular topics, knowledge gaps
• Multi-Tenant Support: Organization-level data isolation
• Real-Time Updates: Incremental indexing, document versioning

Tech Stack

• LlamaIndex: Data indexing, query engines, RAG
• Gemini: GPT-4 for query understanding and generation
• FastAPI: REST API backend
• Streamlit: Web UI for querying and management
• ChromaDB/Pinecone: Vector storage
• PostgreSQL: Metadata and analytics storage
• Redis: Query caching

Quick Start

1. Install Dependencies

cd enterprise_knowledge_base
pip install -r requirements.txt

2. Set Up Environment

Create a .env file in the project root:

# Copy the example
cp .env.example .env

# Edit .env and add your API keys
GEMINI_API_KEY=your_GEMINI_API_KEY_here
GEMINI_MODEL=gemini-3-flash-preview

3. Run Backend API

In one terminal:

cd enterprise_knowledge_base
python -m uvicorn backend.api.main:app --reload --port 8000

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

4. Run Frontend

In another terminal:

cd enterprise_knowledge_base
streamlit run frontend/app.py --server.port 8501

The UI will be available at http://localhost:8501

5. Test the System

1. Open the Streamlit UI at http://localhost:8501
2. Go to "Ingest Documents" tab
3. Upload a PDF or document
4. Go to "Query" tab
5. Ask a question about your document

Architecture

User Query
    ↓
[Query Router] → Determines query type
    ↓
    ├─→ [Document Query Engine] → RAG over documents
    ├─→ [SQL Query Engine] → Natural language to SQL
    ├─→ [API Query Engine] → Query external APIs
    └─→ [Hybrid Query Engine] → Combines multiple sources
    ↓
[Response Synthesizer] → Combines results
    ↓
[Citation Generator] → Adds source citations
    ↓
Response + Sources

Project Structure

enterprise_knowledge_base/
├── backend/
│   ├── core/           # Core LlamaIndex setup
│   ├── engines/        # Query engines
│   ├── ingestion/      # Data ingestion
│   ├── api/            # FastAPI endpoints
│   ├── models/         # Database models
│   └── utils/          # Utilities
├── frontend/           # Streamlit UI
├── data/               # Data storage
└── tests/              # Tests

Usage Examples

Query Documents

from backend.core.knowledge_base import KnowledgeBase

kb = KnowledgeBase()
response = kb.query("What are the key features of our product?")
print(response)

Ingest Documents

kb.ingest_document("path/to/document.pdf", organization_id="org_123")

Query with SQL Generation

response = kb.query_sql("What are the top 10 customers by revenue?")

Tech Stack

• LlamaIndex: Data indexing, query engines, RAG
• FastAPI: REST API backend
• Streamlit: Web UI for querying and management
• ChromaDB/Pinecone: Vector storage
• PostgreSQL: Metadata and analytics storage
• Redis: Query caching

Use Cases

• Enterprise Knowledge Management: Centralized knowledge base for organizations
• Document Q&A: Natural language querying of documents
• Data Integration: Query across multiple data sources
• RAG Applications: Retrieval-augmented generation systems
• Multi-Tenant Knowledge: Organization-level data isolation

Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

License

This project is licensed under the MIT License - see the LICENSE file for details.