🧠 AI Company Brain

An Enterprise GraphRAG Knowledge Platform for Explainable Organizational Intelligence

---

📖 Overview

AI Company Brain is an enterprise-grade GraphRAG (Graph Retrieval-Augmented Generation) knowledge platform designed to transform fragmented organizational information into an intelligent, explainable, and queryable knowledge system.

Traditional enterprise search systems fail because they treat documents as isolated pieces of information. AI Company Brain goes beyond keyword search by combining:

• Semantic vector retrieval
• Knowledge graph reasoning
• Large language models
• Source citations
• Explainable AI workflows

The result is a system capable of answering complex organizational questions while providing transparent reasoning paths and verifiable sources.

---

🚨 The Problem

Modern organizations store information across dozens of disconnected systems:

• Slack conversations
• Internal documentation
• PDFs
• Incident reports
• Notion pages
• GitHub repositories
• Knowledge bases
• Engineering runbooks

Existing solutions suffer from several limitations:

Problem	Existing Systems
Keyword-based search	❌
No semantic understanding	❌
No relationship reasoning	❌
Hallucinated answers	❌
No source attribution	❌
No explainability	❌

This creates a significant knowledge retrieval bottleneck where employees spend more time searching for information than using it.

---

💡 Our Solution

AI Company Brain combines:

Semantic Search
        +
Knowledge Graphs
        +
Graph Traversal
        +
LLM Reasoning
        +
Citation Generation

to create an explainable enterprise intelligence system.

Example query:

"What caused the payment outage?"

Instead of returning documents, the system:

Question
    ↓
Embedding Generation
    ↓
Semantic Search
    ↓
Graph Expansion
    ↓
Knowledge Retrieval
    ↓
LLM Reasoning
    ↓
Source Attribution
    ↓
Final Answer

Result:

• Answer
• Reasoning steps
• Supporting entities
• Supporting documents
• Knowledge graph paths
• Source citations

---

🏗️ System Architecture

                    ┌──────────────────┐
                    │ React Frontend   │
                    └────────┬─────────┘
                             │
                             ▼
                    ┌──────────────────┐
                    │ FastAPI Backend  │
                    └────────┬─────────┘
                             │
         ┌───────────────────┼───────────────────┐
         ▼                   ▼                   ▼

 ┌──────────────┐    ┌──────────────┐    ┌──────────────┐
 │ Embeddings   │    │ Qdrant       │    │ Neo4j        │
 │ Generation   │───▶│ Vector DB    │    │ Knowledge DB │
 └──────────────┘    └──────────────┘    └──────────────┘
                             │                   │
                             ▼                   ▼
                      ┌───────────────────────────┐
                      │ Graph Expansion Engine    │
                      └─────────────┬─────────────┘
                                    ▼
                      ┌───────────────────────────┐
                      │ Gemini 2.5 Flash          │
                      │ Reasoning Engine          │
                      └─────────────┬─────────────┘
                                    ▼
                      ┌───────────────────────────┐
                      │ Citation Engine           │
                      └───────────────────────────┘

---

🚀 Features

🔍 Semantic Search

• Sentence-transformer embeddings
• Dense vector retrieval
• Similarity search
• Context-aware retrieval
• Semantic ranking

---

🕸️ Knowledge Graph

• Entity extraction
• Relationship mapping
• Multi-hop graph traversal
• Explainable graph reasoning
• Interactive graph visualization

---

🧠 GraphRAG

• Semantic retrieval
• Graph expansion
• Knowledge enrichment
• Multi-source reasoning
• Structured context generation

---

🤖 LLM Reasoning

• Gemini 2.5 Flash integration
• Context-aware synthesis
• Multi-document reasoning
• Graph-assisted generation

---

📑 Citation Engine

Every generated answer includes:

• Source documents
• Supporting entities
• Supporting relationships
• Evidence tracing
• Provenance mapping

---

📊 Enterprise Dashboard

• Knowledge statistics
• Database health
• Retrieval analytics
• Graph metrics
• System monitoring

---

🎨 Interactive Frontend

• Real-time chat
• Knowledge graph explorer
• Architecture visualization
• Source inspection
• Responsive UI

---

🧩 Tech Stack

Frontend

Technology	Purpose
React	User Interface
Vite	Build Tool
Tailwind CSS	Styling
JavaScript	Frontend Logic

---

Backend

Technology	Purpose
Python	Core Language
FastAPI	API Framework
Pydantic	Validation
Uvicorn	ASGI Server

---

AI & ML

Technology	Purpose
Sentence Transformers	Embeddings
all-MiniLM-L6-v2	Embedding Model
Gemini 2.5 Flash	LLM Reasoning
GraphRAG	Retrieval Architecture

---

Databases

Technology	Purpose
Qdrant	Vector Database
Neo4j	Graph Database

---

Infrastructure

Technology	Purpose
Docker	Containerization
Docker Desktop	Local Runtime
WSL2	Linux Compatibility

---

⭐ What Makes This Different?

Traditional RAG

Question
    ↓
Vector Search
    ↓
LLM
    ↓
Answer

Problems:

• Limited context
• Weak reasoning
• Hallucinations
• No relationships
• Poor explainability

---

AI Company Brain

Question
     ↓
Embedding Search
     ↓
Vector Retrieval
     ↓
Knowledge Graph
     ↓
Graph Expansion
     ↓
LLM Reasoning
     ↓
Citation Engine
     ↓
Explainable Answer

Advantages:

✅ Semantic understanding

✅ Relationship reasoning

✅ Multi-hop retrieval

✅ Explainable outputs

✅ Source citations

✅ Knowledge graph exploration

---

📷 Demo Screenshots

Login

---

Dashboard

---

AI Assistant

---

Knowledge Graph

---

📂 Project Structure

CodeSmiths/

├── src/                     # React frontend
├── backend/
│   ├── app/
│   ├── embeddings/
│   ├── retrieval/
│   ├── graph/
│   ├── rag/
│   ├── citations/
│   ├── evaluation/
│   └── brain/
│
├── data/
├── agent_orchestration/
│
├── test_embeddings.py
├── test_qdrant.py
├── test_neo4j.py
├── test_graph_expansion.py
├── test_graph_rag.py
└── test_citations.py

---

⚙️ System Requirements

Minimum:

• Windows 10/11
• Linux
• macOS
• 8 GB RAM
• 15 GB free storage

Recommended:

• 16 GB RAM
• Quad-core CPU
• SSD
• Docker Desktop

---

🛠️ Prerequisites

Install:

• Python 3.12+
• Node.js 20+
• npm
• Git
• Docker Desktop
• WSL2 (Windows)

---

🐳 Start Databases

Qdrant

docker run -d \
--name qdrant \
-p 6333:6333 \
-p 6334:6334 \
qdrant/qdrant

---

Neo4j

docker run -d \
--name neo4j \
-p 7474:7474 \
-p 7687:7687 \
-e NEO4J_AUTH=neo4j/password123 \
neo4j

---

📥 Installation

Clone repository:

git clone https://github.com/YOUR_USERNAME/CodeSmiths.git

cd CodeSmiths

---

Backend Setup

pip install -r requirements.txt

or

pip install \
fastapi \
uvicorn \
sentence-transformers \
qdrant-client \
neo4j \
google-generativeai

---

Create:

.env

Add:

GEMINI_API_KEY=YOUR_GEMINI_API_KEY

---

Frontend Setup

npm install

---

🚀 Running the Project

Start Backend

python demo_server.py

Backend:

http://localhost:8000

API Docs:

http://localhost:8000/docs

---

Start Frontend

npm run dev

Frontend:

http://localhost:5173

---

🧪 Testing

Run individual subsystem tests:

python test_qdrant.py
python test_neo4j.py
python test_graph_expansion.py
python test_graph_rag.py
python test_citations.py

---

📈 Example Query

Input:

What caused the payment outage?

Output:

Answer:
The payment outage was caused by the Redis Cluster outage.

Sources:
[1] payment_incident.md
[2] redis_outage.md
[3] slack_thread.md

Graph Entities:
• Payment Service
• Incident #1001
• Slack Thread
• Redis Cluster

---

🔮 Future Improvements

• Multi-user authentication
• Role-based access control
• Real document connectors
• Slack integration
• Notion integration
• GitHub integration
• Hybrid retrieval
• Agentic workflows
• Real-time graph updates
• Fine-tuned retrieval models
• Observability dashboard
• Production deployment

---

👥 Team

Built by Team CodeSmiths.

Contributors:

• Backend Ingestion
• Frontend Dashboard
• Agent Orchestration
• AI Company Brain / GraphRAG

---

📜 License

This project is intended for educational, research, and demonstration purposes.

---

"From fragmented information to organizational intelligence."