title	Agents Knowledge Base
emoji	🤖
colorFrom	blue
colorTo	purple
sdk	docker
pinned	false
tags	agent rag knowledge-base customer-support groq chromadb streamlit llm python

🤖 Agents Knowledge Base — Customer Support Agent

AI-powered customer support agent with two tools: search a pre-indexed knowledge base and escalate to human support when needed. Built with Groq + ChromaDB + Streamlit.

---

✨ Features

• 🔍 Knowledge base search — semantic search over pre-indexed PDFs
• 🚨 Human escalation — automatic escalation with ticket creation
• 💬 Chat interface — conversation history with Streamlit
• 📋 Escalation dashboard — view pending tickets in the sidebar
• ⚡ Ultra-fast — powered by Groq inference

---

🧠 How It Works

Customer question
       │
       ▼
Escalation keywords detected?
   YES → Create ticket → Human agent
   NO  → Search knowledge base
              │
        Found answer?
          YES → Answer customer
          NO  → Create ticket → Human agent

Escalation triggers :

• Customer explicitly requests a human agent
• Keywords detected: urgent, complaint, frustrated...
• No relevant answer found in the knowledge base

---

📦 Tech Stack

Component	Tool	Role
LLM	Groq (llama-3.3-70b-versatile)	Answer generation
Embeddings	HuggingFace (all-MiniLM-L6-v2)	Text vectorization
Vector Store	ChromaDB	Semantic search
UI	Streamlit	Chat interface

---

🚀 Getting Started (Local)

Prerequisites

• Python 3.11+
• A Groq API key (free)

1. Clone the repository

git clone https://github.com/salmazenn/agents-knowledge-bas.git
cd agents-knowledge-bas

2. Create a virtual environment

python3 -m venv .venv-agents
source .venv-agents/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Set up environment variables

cp env.example .env

Edit .env:

# Required
GROQ_API_KEY=your_groq_api_key_here

# Optional
GROQ_MODEL=llama-3.3-70b-versatile
CHROMA_PERSIST_DIR=./data/chroma
DOCS_DIR=./docs
CHUNK_SIZE=1000
CHUNK_OVERLAP=100
TOP_K=3
ESCALATION_LOG=./data/escalations.json

5. Add your documents

cp your-faq.pdf docs/

6. Index your documents

python src/ingest.py

7. Launch the app

streamlit run src/app.py

Open http://localhost:8501 🎉

---

⚙️ Configuration Parameters

Variable	Default	Description
GROQ_API_KEY	—	Required. Your Groq API key
GROQ_MODEL	llama-3.3-70b-versatile	Groq model
CHROMA_PERSIST_DIR	./data/chroma	ChromaDB storage path
DOCS_DIR	./docs	Folder with your PDFs
CHUNK_SIZE	1000	Chunk size in tokens
CHUNK_OVERLAP	100	Overlap between chunks
TOP_K	3	Number of chunks retrieved
ESCALATION_LOG	./data/escalations.json	Escalation tickets log

---

📁 Project Structure

agents-knowledge-base/
├── src/
│   ├── ingest.py       # Index PDFs into ChromaDB
│   ├── agent.py        # Agent logic (search + escalation)
│   └── app.py          # Streamlit chat interface
├── docs/               # Drop your PDFs here
├── data/
│   ├── chroma/         # ChromaDB (auto-generated)
│   └── escalations.json # Escalation tickets
├── .env                # API keys (never commit!)
├── env.example         # Environment template
├── requirements.txt
├── Dockerfile
└── README.md

---

🔭 Roadmap

• Email notifications for escalation tickets
• Admin dashboard to manage tickets
• Multi-language support
• Integration with CRM (Salesforce, HubSpot)
• Analytics dashboard

---

📄 License

MIT — free to use, modify and distribute.