This repo has been updated to use LangGraph instead of legacy LangChain's AgentExecutor. Additionally, Ollama is used instead of OpenAI. [Dec 4, 2024].
"Tutorial on building customer chatbot for coffee shop with Agent LLM and RAG". This is a repo that we demo on our YouTube video Tutorial - Build a Customer Contact Chatbot with Gen-AI: LangChain, Chroma, & Gradio
- Embark on this journey with us to explore the potential of Generative AI in enhancing customer service through an intelligent chatbot. Let's build, learn, and innovate together!
- This step-by-step tutorial is ideal for software and AI developers, forward-thinking business managers, and tech enthusiasts, this video guides you through creating a demo of an intelligent chatbot for a mock-up coffee shop scenario, from inception to a fully functional demo.
- Project Setup: Developing a chatbot that provides store information, coffee product details, and helps customers choose the right coffee beans. We do that using text and CSV files.
- Technology Insights: Agent-based workflow with LLM and RAG (Retrieval Augmented Generation).
- Coding Session: Hands-on walkthrough using Python, featuring LangChain for its LLM prowess, Chroma for information storage, and Gradio for a quich chatbot interface for a demo.
- Demo Showcase: Experience the chatbot in action, demonstrating its capability to manage customer interactions, perform small talks, and handle product inquiries seamlessly.
👍 Like | 🔗 Share | 📢 Subscribe
Follow us on YouTube, LinkedIn, and Facebook! Look for @casedonebyai
💬 Comments? Questions? We value your feedback and look forward to engaging with you!
0:10 Scenario: Mock-up Coffee Shop
0:40 RAG review and quick intro
1:52 Coding session starts from here
2:00 Documents needed for RAG
7:23 RAG Step 1: Indexing and Saving to Chroma index
14:43 RAG Step 2: Retrieval, Loading Chroma index, and Using Retriever
21:26 In LangChain, Build LLM-based Agent and Register RAG tool in LangChain
28:39 Launch and test chatbot demo interface with Gradio!
37:56 Briefly seeing 'verbose' from LLM agent action
39:20 Wrapping up and summary
- Make sure you use the appropriate environment. You can install modules using
requirements.txt. - If you find an error about OpenAI, it could be that you need to specifcy OpenAI API key. Make a file called
openai_api_keyinsecretfolder and keep your key there. - Ollama with Llama3.2-3B should be running. You can set up by:
ollama pull llama3.2
ollama serve