Skip to main content

Search-Enhanced Chat Interface

Presenter​

Colin McNamara is Engineer 🔧 | Entrepreneur 💼 | Sustainability Champion 🌱 | Supply Chain Strategist 🔗 | Private Label Pioneer 🏷️ | Build a Better Tomorrow 🌍 Currently at Always Cool Brands in Austin, TX.

"Our focus is really low-stress learning and sharing. We're not trying to be experts. We're all learning. This is a fast-moving project. We are here to connect with other early adopters of AI middleware, and specifically, focused around the LangChain project."

Connect with Colin:

Workshop Details​

What to Expect​

During this session, you'll discover how to harness the power of ChatGPT models via API access and customize a web interface. Rest assured, your data going through this API will not be used for training.

Requirements​

  • A working computer.
  • A free Google Colab account (Google login) for Jupyter notebooks
  • A working paid OpenAI API key - https://platform.openai.com/account/api-keys
  • A can-do attitude and an eagerness to explore this exciting new form of software
  • If you prefer to attend the lecture and soak in the concepts without participating in the lab, that's perfectly fine too!
  • This event is part of our multi-part lecture and lab series designed to equip you with the skills to set up your own AI microservices centered around LangChain.

Lab Overview​

Learn how to build a chat interface with integrated search capabilities, combining conversation and document retrieval functionalities using LangChain. This lab demonstrates how to create an AI assistant that can search the internet in real-time to provide up-to-date information while maintaining conversation context.

Lab Materials​

Key Components​

  • Agent System:

    from langchain.agents import ConversationalChatAgent, AgentExecutor
    from langchain.callbacks import StreamlitCallbackHandler
    from langchain.chat_models import ChatOpenAI
    from langchain.memory import ConversationBufferMemory
    from langchain.memory.chat_message_histories import StreamlitChatMessageHistory
    from langchain.tools import DuckDuckGoSearchRun

  • Memory Management:

    • StreamlitChatMessageHistory for UI integration
    • ConversationBufferMemory for context retention
    • Message history reset functionality

Technical Implementation​

Agent Configuration​

llm = ChatOpenAI(
    model_name="gpt-3.5-turbo",
    openai_api_key=openai_api_key,
    streaming=True
)
tools = [DuckDuckGoSearchRun(name="Search")]
chat_agent = ConversationalChatAgent.from_llm_and_tools(
    llm=llm,
    tools=tools
)

Executor Setup​

executor = AgentExecutor.from_agent_and_tools(
    agent=chat_agent,
    tools=tools,
    memory=memory,
    return_intermediate_steps=True,
    handle_parsing_errors=True
)

Features​

  • Interactive chat interface
  • Real-time web search integration
  • Conversation history management
  • Intermediate step visualization
  • Error handling and recovery
  • Streaming responses
  • Status updates for search operations

Implementation Steps​

  1. Environment Setup:

    pip install langchain openai streamlit duckduckgo-search

  2. Memory Configuration:

    • Initialize StreamlitChatMessageHistory
    • Configure ConversationBufferMemory
    • Set up message history management

  3. Agent Setup:

    • Configure ChatOpenAI model
    • Set up DuckDuckGo search tool
    • Initialize ConversationalChatAgent

  4. Interface Implementation:

    • Create Streamlit chat interface
    • Implement message display
    • Add search status indicators
    • Configure streaming callbacks

Technical Components​

  • ChatOpenAI for language model interaction
  • DuckDuckGoSearchRun for web search
  • StreamlitCallbackHandler for UI updates
  • ConversationBufferMemory for context
  • AgentExecutor for tool orchestration

Advanced Features​

  • Intermediate step visualization
  • Search operation status tracking
  • Error handling and recovery
  • Message history persistence
  • Real-time response streaming
  • Tool execution logging

Best Practices​

  • Initialize chat history properly
  • Handle API key securely
  • Implement proper error handling
  • Display search progress
  • Maintain conversation context
  • Provide clear status updates

Prerequisites​

  • Google Colab account
  • OpenAI API key
  • Basic Python knowledge
  • Understanding of:
    • Conversational AI
    • Search systems
    • Web interfaces
    • Agent-based systems

Implementation Details​

  1. Message Management:

    • Store messages in StreamlitChatMessageHistory
    • Display with appropriate avatars
    • Track intermediate steps
    • Handle message history reset

  2. Search Integration:

    • Configure DuckDuckGo search tool
    • Display search status
    • Show search results
    • Handle search errors

  3. Agent Execution:

    • Process user input
    • Execute search operations
    • Generate responses
    • Update conversation history

Resources​