🤖 LocalLoop – Smart Commerce Agent

AI Agent for Shopping, Travel, and Food

LocalLoop is a multilingual AI assistant built to deliver intelligent, contextual, and personalized experiences across commercial domains such as food, travel, and marketplaces. Leveraging RAG (Retrieval-Augmented Generation), knowledge graphs, and modular agent orchestration, LocalLoop enables users to interact via text or voice in both English 🇬🇧 and Indonesian 🇮🇩.

---

✨ Key Features

• 🍽️ Food Agent – Discover restaurants, browse menus, and place orders
• 🛫 Travel Agent – Search for flights, reserve hotels, and plan itineraries
• 🛍️ Marketplace Agent – Buy or sell new and second-hand items
• 🧠 Knowledge Graph Profile – Persistent and reusable user personalization
• 🗣️ Multimodal Interaction – Text and voice input support
• 🌐 Multilingual Support – English & Indonesian

---

🧠 AI Architecture & Agent Design

🧩 Modular Agent Orchestration

• Natural language-based intent & domain detection
• Dynamic fallback logic tailored per domain
• RAG-powered retrieval with FAISS or Weaviate for contextual relevance

🔎 Domain Agents

• FoodAgent, TravelAgent, MarketplaceAgent
• Manage slot filling, session memory, task execution, and fallbacks

🧬 Knowledge Graph Personalization

• RDF-based graph representing user preferences
• Built incrementally through in-app interactions
• Visualizable through a UI (graph.html)

---

⚙️ Tech Stack

Component	Description
Groq + LLaMA 3.1	High-speed, low-latency LLM inference
FAISS / Weaviate	Vector-based retrieval for RAG
SERP API / Tavily	Fallback web search
Infobip	Voice and WhatsApp integration
LangChain	Orchestration, memory, and chaining
Flask	Lightweight web backend

---

✅ Challenge Criteria Fulfilled

• ✅ Food ordering
• ✅ Travel booking
• ✅ Product marketplace
• ✅ Voice-first & multimodal interface
• ✅ Knowledge graph-based user profile
• ✅ AI-powered recommendations
• ✅ Reusable personalization across domains
• ✅ Live demo–ready
• ✅ Modular and scalable architecture

---

🧪 Getting Started

1. Clone the Repository

git clone https://github.com/bambangirawans/localloop.git
cd localloop

2. Install Dependencies

pip install -r requirements.txt

3. Configure Environment Variables

cp .env.example .env

Then, add the following API keys to your .env file:

• GROQ_API_KEY – for LLaMA 3.1 via Groq
• SERP_API_KEY or TAVILY_API_KEY – for fallback search
• INFOBIP_KEY – for WhatsApp and voice integration

4. (Optional) Build FAISS Index

python scripts/build_faiss_index.py

5. Run the Application

python run.py

---

🌐 User Interface

File	Description
index.html	Main chat interface
graph.html	RDF user profile graph visualizer
voice_handler.py	Voice input and speech recognition logic

---

🧰 Useful Scripts

• scripts/build_faiss_index.py – Builds FAISS index from data
• scripts/generate_rag_data_with_llm.py – Auto-generates RAG knowledge base
• scripts/visualize_graph.py – Visualizes the user profile graph

---

🤝 Contributing

We welcome feedback, ideas, and contributions!

• 🐞 Please open an issue or pull request to contribute

---

🔖 License

MIT License © 2025 — LocalLoop Team