🏢 YSS — Internal Knowledge Base Chatbot

A secure, RAG-powered internal knowledge base chatbot built with Python and Streamlit. Employees can ask natural language questions about company policies, HR documents, IT guides, and sales materials. Admins can upload documents through a password-protected interface.

---

Features

• RAG-powered answers — retrieves relevant document excerpts before answering; no hallucinations
• Multi-format ingestion — PDF, DOCX, TXT, Markdown
• Admin upload panel — password-protected, categorized document management
• Knowledge base browser — view all indexed docs, test retrieval queries
• Streaming responses — answers appear token by token like a real chatbot
• Source citations — every answer cites which documents were used
• Conversation memory — maintains context across a session (last 10 turns)
• Docker-ready — one command to deploy on any internal cloud

---

Tech Stack

Layer	Tool
Frontend	Streamlit
LLM	Anthropic Claude (claude-sonnet)
Vector Search	TF-IDF + Cosine Similarity (built-in, no GPU needed)
Document Parsing	PyPDF2, python-docx
Deployment	Docker + Docker Compose

---

Quick Start (Local)

1. Clone and enter project

git clone <your-repo-url>
cd company-kb

2. Create virtual environment

python -m venv venv
source venv/bin/activate        # macOS/Linux
# venv\Scripts\Activate.ps1    # Windows PowerShell

3. Install dependencies

pip install -r requirements.txt

4. Configure environment

cp .env.example .env

Edit .env and set your ANTHROPIC_API_KEY. Get one at https://console.anthropic.com

# .env
ANTHROPIC_API_KEY=sk-ant-...
ADMIN_PASSWORD=your-secure-password

5. Load the .env variables

# macOS/Linux
export $(cat .env | xargs)

# Windows PowerShell
Get-Content .env | ForEach-Object { $k,$v = $_ -split '=',2; [System.Environment]::SetEnvironmentVariable($k,$v) }

6. Seed sample documents (optional)

python seed_sample_docs.py

7. Run the app

streamlit run app.py

Open http://localhost:8501

---

Uploading Documents

1. Go to Admin Upload in the sidebar
2. Enter the admin password (default: admin123 — change this in .env)
3. Select a category, drag and drop files, click Ingest
4. Documents are immediately searchable in the chat

Supported formats: PDF, DOCX, TXT, Markdown

---

Docker Deployment (Internal Cloud)

Build and run

# Set your API key
echo "ANTHROPIC_API_KEY=sk-ant-..." > .env
echo "ADMIN_PASSWORD=your-secure-password" >> .env

# Build and start
docker-compose up -d

Access at http://your-server-ip:8501

Useful Docker commands

docker-compose logs -f          # View logs
docker-compose down             # Stop
docker-compose up -d --build    # Rebuild after code changes
docker-compose ps               # Check status

Documents and the vector store are persisted in Docker volumes — safe across restarts.

---

Project Structure

company-kb/
├── app.py                          # Main chat interface
├── pages/
│   ├── 1_Admin_Upload.py           # Document upload (password-protected)
│   └── 2_Knowledge_Base.py         # Browse docs + test search
├── utils/
│   ├── rag_engine.py               # RAG core: chunking, TF-IDF, retrieval
│   └── llm_client.py               # Anthropic Claude API wrapper
├── sample_docs/                    # Sample documents for demo
├── documents/                      # Uploaded documents (auto-created)
├── vector_store/                   # Index + metadata (auto-created)
├── .streamlit/config.toml          # Streamlit theme config
├── seed_sample_docs.py             # One-time sample data loader
├── requirements.txt
├── Dockerfile
├── docker-compose.yml
└── .env.example

---

Upgrading to Semantic Search (Optional)

The current engine uses TF-IDF which works well and requires no GPU. To upgrade to semantic embeddings (better for conceptual questions):

1. Uncomment sentence-transformers in requirements.txt
2. In rag_engine.py, replace the TFIDFStore with a sentence-transformers + faiss implementation

---

Security Notes

• Change ADMIN_PASSWORD from the default before deploying
• Keep ANTHROPIC_API_KEY secret — never commit .env to version control
• For production, add SSO/LDAP authentication in front of Streamlit
• The Docker setup runs on an internal port — do not expose directly to the internet without a reverse proxy (nginx) and HTTPS

---

License

Internal use only.