Hands-on practical: Build a Simple Q&A Bot

Let's put the concepts from this chapter into practice by building a functional, albeit simple, question-answering (Q&A) bot. This exercise will solidify your understanding of making API calls, sending prompts, and processing responses programmatically. We'll create a command-line application that takes a user's question, sends it to an LLM via its API, and displays the answer.

Prerequisites

Before we begin, ensure you have the following set up:

1. Python Environment: A working Python 3 installation.
2. LLM API Access: An API key from an LLM provider (e.g., OpenAI, Anthropic, Google). For this example, we'll structure the code assuming the OpenAI API, but the principles are similar for others.
3. Required Libraries: You'll need a library to make HTTP requests or a provider-specific library. We'll use the openai library for convenience. Install it using pip:

   pip install openai
   

4. API Key Security: Never hardcode your API key directly in your script. Use environment variables. Set an environment variable named OPENAI_API_KEY with your actual key.

• On Linux/macOS: export OPENAI_API_KEY='your_api_key_here'
• On Windows (Command Prompt): set OPENAI_API_KEY=your_api_key_here
• On Windows (PowerShell): $env:OPENAI_API_KEY='your_api_key_here'

Core Logic of the Q&A Bot

Our bot will follow these steps in a loop:

1. Prompt the User: Ask the user to enter a question.
2. Get Input: Read the question from the command line.
3. Check for Exit: If the user types "quit", "exit", or similar, terminate the program.
4. Construct Prompt: Create a simple prompt incorporating the user's question. We'll instruct the LLM to act as a helpful assistant.
5. Call LLM API: Send the prompt to the LLM API, specifying the model and any relevant parameters (like temperature or max_tokens).
6. Process Response: Extract the answer text from the API response.
7. Handle Errors: Include basic error handling for API calls (e.g., network issues, authentication problems, rate limits).
8. Display Answer: Print the LLM's response to the console.
9. Loop: Repeat the process.

Python Implementation

Let's write the Python code. Create a file named qa_bot.py.

import os
import openai
from openai import OpenAI # Use the updated OpenAI library structure

# --- Configuration ---
# Attempt to load the API key from an environment variable
try:
    client = OpenAI(api_key=os.environ.get("OPENAI_API_KEY"))
except openai.OpenAIError as e:
    print(f"Error initializing OpenAI client: {e}")
    print("Please ensure the OPENAI_API_KEY environment variable is set.")
    exit()

MODEL_NAME = "gpt-3.5-turbo" # Or choose another suitable model

# --- Core Function ---
def get_llm_response(user_question):
    """
    Sends the user's question to the LLM API and returns the response.
    """
    system_prompt = "You are a helpful assistant. Answer the user's question clearly and concisely."
    
    try:
        completion = client.chat.completions.create(
            model=MODEL_NAME,
            messages=[
                {"role": "system", "content": system_prompt},
                {"role": "user", "content": user_question}
            ],
            temperature=0.7, # Adjust for creativity vs. determinism
            max_tokens=150   # Limit the response length
        )
        
        # Extracting the response text
        # Note: The exact structure might vary slightly based on API version updates
        if completion.choices and len(completion.choices) > 0:
            response_text = completion.choices[0].message.content.strip()
            return response_text
        else:
            return "Error: No response received from the API."
            
    except openai.APIConnectionError as e:
        return f"API Connection Error: {e}"
    except openai.RateLimitError as e:
        return f"API Rate Limit Exceeded: {e}"
    except openai.AuthenticationError as e:
        return f"API Authentication Error: {e}. Check your API key."
    except openai.APIError as e:
        return f"API Error: {e}"
    except Exception as e:
        # Catch any other unexpected errors
        return f"An unexpected error occurred: {e}"

# --- Main Interaction Loop ---
def main():
    print("Simple Q&A Bot (type 'quit' or 'exit' to stop)")
    print("-" * 30)
    
    while True:
        user_input = input("You: ")
        
        if user_input.lower() in ["quit", "exit"]:
            print("Bot: Goodbye!")
            break
            
        if not user_input:
            continue
            
        print("Bot: Thinking...") # Provide feedback while waiting
        
        llm_answer = get_llm_response(user_input)
        
        print(f"Bot: {llm_answer}")
        print("-" * 30)

if __name__ == "__main__":
    main()

Code Explanation

1. Imports: We import os to access environment variables and openai for the API interaction.
2. Configuration: We initialize the OpenAI client. The API key is fetched from the OPENAI_API_KEY environment variable. Basic error handling is included here to catch initialization problems. We also define the MODEL_NAME.
3. get_llm_response(user_question) Function:
   • Takes the user_question as input.
   • Defines a system_prompt to provide context or instructions to the LLM (in this case, setting its role).
   • Uses client.chat.completions.create to call the API. This is the standard method for chat-based models like GPT-3.5 Turbo or GPT-4.
   • The messages parameter follows the chat format, including roles ("system", "user").
   • temperature and max_tokens are included as examples of parameters discussed earlier in the chapter. temperature=0.7 allows for some variability in responses, while max_tokens=150 limits the length to avoid overly long answers.
   • Response Parsing: It accesses the response content via completion.choices[0].message.content. The structure of the response object is important here; you might need to inspect the actual response object (print(completion)) during development if using different libraries or models.
   • Error Handling: try...except blocks catch specific openai exceptions (like connection errors, rate limits, authentication issues) as well as general exceptions, returning informative error messages instead of crashing the bot.
4. main() Function:
   • Prints a message.
   • Enters an infinite while True loop for continuous interaction.
   • input("You: ") prompts the user and reads their input.
   • Checks for "quit" or "exit" commands to break the loop.
   • Calls get_llm_response with the user's input.
   • Prints the returned answer (or error message) prefixed with "Bot: ".
5. if __name__ == "__main__":: This standard Python construct ensures that the main() function is called only when the script is executed directly.

Running the Bot

1. Save the code as qa_bot.py.
2. Make sure your OPENAI_API_KEY environment variable is set in your terminal session.
3. Run the script from your terminal:

   python qa_bot.py
   

Example Interaction

Simple Q&A Bot (type 'quit' or 'exit' to stop)
------------------------------
You: What is the capital of France?
Bot: Thinking...
Bot: The capital of France is Paris.
------------------------------
You: Explain the concept of temperature in LLM APIs.
Bot: Thinking...
Bot: In LLM APIs, the 'temperature' parameter controls the randomness of the output. A lower temperature (e.g., 0.2) makes the output more focused and deterministic, often choosing the most likely next word. A higher temperature (e.g., 0.8) increases randomness, leading to more creative or diverse responses, but potentially less coherent ones. It essentially adjusts the probability distribution of the next token prediction.
------------------------------
You: quit
Bot: Goodbye!

Next Steps and Improvements

This bot is basic, but it demonstrates the core interaction pattern. Here are some ways you could extend it:

• Conversation History (Memory): Modify the get_llm_response function to accept and include previous turns of the conversation in the messages list. This allows the LLM to have context about what was discussed earlier. (Covered in later chapters on frameworks).
• Streaming Responses: Implement response streaming (if the API supports it) to display the answer word by word as it's generated, improving perceived responsiveness. (Covered earlier in this chapter).
• Different Prompting Techniques: Experiment with more advanced prompts from Chapter 2 (e.g., role prompting, asking for specific output formats).
• Parameter Tuning: Adjust temperature, max_tokens, top_p, etc., to see how they affect the output quality and style for different types of questions.
• Model Selection: Try different models available through the API.
• User Interface: Build a simple web interface (using Flask or Streamlit) instead of a command-line interaction.

This hands-on exercise provides a tangible example of how the API interaction concepts covered in this chapter translate into a working application. By successfully building and running this bot, you've taken a significant step towards integrating LLMs into your own software projects.