Module 2: OpenAI API - Introduction

Mini-Project: Technical Support Chatbot

Project overview

This is the integrative project for Module 2. You'll apply EVERYTHING you've learned to build a production-ready technical support chatbot.

Features:

✅ Conversations with context (remembers history)
✅ Custom system message (specific behavior)
✅ Optimized parameters (temperature, max_tokens)
✅ Robust error handling (retries, logging)
✅ Cost tracking (monitors spending)
✅ Persistence (saves conversations to JSON)

Time: 60-90 minutes
Difficulty: Medium-High

🎯 Project goal

Build a CLI chatbot that:

Answers technical support questions
Maintains conversational context
Handles errors gracefully (doesn't crash)
Tracks cost per conversation
Saves logs for analysis

📋 Specifications

Functionality:

Chatbot role:

Technical support assistant for the "TechApp" app
Answers FAQs about password reset, billing, features
Escalates to a human if it can't resolve the issue

Special commands:

salir / exit → Ends the conversation
historial → Shows the full history
costo → Shows the accumulated cost

Persistence:

Saves each conversation to conversations/conversation_TIMESTAMP.json
Error logs in logs/errors.log

💻 Implementation

Project structure:

chatbot-soporte/
├── .env                    # API key
├── .gitignore             # Prevents leaks
├── chatbot.py             # Main code
├── conversations/         # Saved conversations
│   └── conversation_20240215_103045.json
└── logs/
    └── errors.log         # Error logs

Full code (`chatbot.py`):

#!/usr/bin/env python3
"""
Technical Support Chatbot - TechApp
Module 2: OpenAI API - Final Project
"""

import os
import json
import time
import logging
from datetime import datetime
from typing import Optional, List, Dict
from pathlib import Path

from dotenv import load_dotenv
from openai import OpenAI, RateLimitError, APIError, APITimeoutError

# ============================================================================
# CONFIGURATION
# ============================================================================

load_dotenv()

# Create directories
Path("conversations").mkdir(exist_ok=True)
Path("logs").mkdir(exist_ok=True)

# Logging
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s',
    handlers=[
        logging.FileHandler('logs/errors.log'),
        logging.StreamHandler()
    ]
)
logger = logging.getLogger(__name__)

# OpenAI client
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

# ============================================================================
# CHATBOT CONFIGURATION
# ============================================================================

SYSTEM_MESSAGE = """
You are a technical support assistant for TechApp, a productivity application.

Your role:
1. Help users with frequently asked questions
2. Be friendly, clear, and concise
3. If you don't know something, say: "Let me escalate this to a human agent"
4. NEVER make up information (prices, dates, features)

Information about TechApp:
- Password reset: Settings > Security > Reset Password
- Billing: $10/month basic plan, $30/month premium plan
- Features: Task management, Calendar, Notes, Multi-device sync
- Support: support@techapp.com

Always respond in English, 4 sentences maximum.
""".strip()

# ============================================================================
# CHATBOT CLASS
# ============================================================================

class TechSupportBot:
    """Technical support chatbot with OpenAI."""
    
    def __init__(self):
        self.messages: List[Dict[str, str]] = [
            {"role": "system", "content": SYSTEM_MESSAGE}
        ]
        self.total_tokens = 0
        self.total_cost = 0.0
        self.conversation_id = datetime.now().strftime("%Y%m%d_%H%M%S")
        
    def chat(self, user_message: str, max_retries: int = 3) -> Optional[str]:
        """
        Send a message and return the response with retry.
        
        Args:
            user_message: The user's message
            max_retries: Maximum attempts if there's an error
            
        Returns:
            The bot's response, or None if it fails
        """
        
        # Add the user's message
        self.messages.append({"role": "user", "content": user_message})
        
        # Retry with exponential backoff
        for attempt in range(max_retries):
            try:
                response = client.chat.completions.create(
                    model="gpt-3.5-turbo",
                    messages=self.messages,
                    temperature=0.3,  # Low (consistent support)
                    max_tokens=150,   # Short answers
                    timeout=30.0
                )
                
                # Extract the response
                assistant_message = response.choices[0].message.content
                self.messages.append({"role": "assistant", "content": assistant_message})
                
                # Update metrics
                self._update_metrics(response.usage)
                
                logger.info(f"✅ Successful request | Tokens: {response.usage.total_tokens}")
                return assistant_message
                
            except RateLimitError:
                logger.warning(f"⚠️ Rate limit | Attempt {attempt + 1}/{max_retries}")
                if attempt == max_retries - 1:
                    logger.error("❌ Persistent rate limit")
                    return None
                time.sleep(2 ** attempt)
                
            except (APIError, APITimeoutError) as e:
                logger.warning(f"⚠️ API error | Attempt {attempt + 1}/{max_retries}")
                if attempt == max_retries - 1:
                    logger.error(f"❌ Persistent error: {e}")
                    return None
                time.sleep(2 ** attempt)
                
            except Exception as e:
                logger.error(f"❌ Unexpected error: {type(e).__name__} | {e}")
                return None
        
        return None
    
    def _update_metrics(self, usage):
        """Update cost and token metrics."""
        self.total_tokens += usage.total_tokens
        
        # GPT-3.5-turbo pricing (Feb 2026)
        input_cost = (usage.prompt_tokens / 1_000_000) * 0.50
        output_cost = (usage.completion_tokens / 1_000_000) * 1.50
        self.total_cost += input_cost + output_cost
    
    def show_history(self):
        """Show the conversation history."""
        print("\n" + "="*60)
        print("CONVERSATION HISTORY")
        print("="*60)
        
        for i, msg in enumerate(self.messages[1:], 1):  # Skip system
            role = "YOU" if msg["role"] == "user" else "BOT"
            print(f"\n[{i}] {role}: {msg['content']}")
        
        print("\n" + "="*60 + "\n")
    
    def show_cost(self):
        """Show the accumulated cost."""
        print(f"\n💰 Accumulated cost: ${self.total_cost:.6f}")
        print(f"📊 Total tokens: {self.total_tokens}")
        print(f"📝 Messages: {len(self.messages) - 1}\n")  # -1 for system
    
    def save_conversation(self):
        """Save the conversation to JSON."""
        filename = f"conversations/conversation_{self.conversation_id}.json"
        
        data = {
            "conversation_id": self.conversation_id,
            "timestamp": datetime.now().isoformat(),
            "messages": self.messages[1:],  # Skip system message
            "total_tokens": self.total_tokens,
            "total_cost": self.total_cost,
            "message_count": len(self.messages) - 1
        }
        
        with open(filename, "w", encoding="utf-8") as f:
            json.dump(data, f, indent=2, ensure_ascii=False)
        
        logger.info(f"💾 Conversation saved: {filename}")
        print(f"💾 Conversation saved: {filename}")

# ============================================================================
# MAIN - CLI INTERFACE
# ============================================================================

def main():
    """The chatbot's main CLI."""
    
    # Banner
    print("\n" + "="*60)
    print("🤖 TECHNICAL SUPPORT CHATBOT - TechApp")
    print("="*60)
    print("\nSpecial commands:")
    print("  - 'salir' / 'exit'  → End the conversation")
    print("  - 'historial'       → See the full history")
    print("  - 'costo'           → See the accumulated cost")
    print("\n" + "="*60 + "\n")
    
    # Create the bot
    bot = TechSupportBot()
    
    # Main loop
    while True:
        try:
            # User input
            user_input = input("You: ").strip()
            
            # Special commands
            if user_input.lower() in ["salir", "exit", "quit"]:
                print("\n👋 Thanks for contacting TechApp! See you soon.\n")
                bot.show_cost()
                bot.save_conversation()
                break
            
            if user_input.lower() == "historial":
                bot.show_history()
                continue
            
            if user_input.lower() == "costo":
                bot.show_cost()
                continue
            
            # Validate input
            if not user_input:
                print("⚠️ Please type a message.\n")
                continue
            
            # Send to the bot
            response = bot.chat(user_input)
            
            if response:
                print(f"\nBot: {response}\n")
            else:
                print("\n❌ Sorry, there was an error. Please try again.\n")
        
        except KeyboardInterrupt:
            print("\n\n👋 Conversation interrupted by the user.\n")
            bot.show_cost()
            bot.save_conversation()
            break
        
        except Exception as e:
            logger.critical(f"❌ Critical error in the main loop: {e}")
            print(f"\n❌ Unexpected error: {e}\n")
            bot.save_conversation()
            break

if __name__ == "__main__":
    # Check the API key
    if not os.getenv("OPENAI_API_KEY"):
        print("❌ ERROR: OPENAI_API_KEY not found in .env")
        exit(1)
    
    main()

🚀 Usage

1. Run it:

python chatbot.py

2. Example conversation:

============================================================
🤖 TECHNICAL SUPPORT CHATBOT - TechApp
============================================================

Special commands:
  - 'salir' / 'exit'  → End the conversation
  - 'historial'       → See the full history
  - 'costo'           → See the accumulated cost

============================================================

You: Hi, how do I reset my password?

Bot: To reset your password in TechApp:
1. Go to Settings
2. Select Security
3. Click Reset Password
4. You'll receive an email with instructions

You: I'm not receiving the email

Bot: If you didn't receive the reset email:
1. Check your Spam/Junk folder
2. Confirm the registered email is correct
3. Wait 5 minutes (there's sometimes a delay)
If it still doesn't arrive, let me escalate this to a human agent.

You: historial

============================================================
CONVERSATION HISTORY
============================================================

[1] YOU: Hi, how do I reset my password?

[2] BOT: To reset your password in TechApp:
1. Go to Settings
2. Select Security
3. Click Reset Password
4. You'll receive an email with instructions

[3] YOU: I'm not receiving the email

[4] BOT: If you didn't receive the reset email:
1. Check your Spam/Junk folder
2. Confirm the registered email is correct
3. Wait 5 minutes (there's sometimes a delay)
If it still doesn't arrive, let me escalate this to a human agent.

============================================================

You: costo

💰 Accumulated cost: $0.000456
📊 Total tokens: 285
📝 Messages: 4

You: salir

👋 Thanks for contacting TechApp! See you soon.

💰 Accumulated cost: $0.000456
📊 Total tokens: 285
📝 Messages: 4

💾 Conversation saved: conversations/conversation_20240215_103045.json

📊 Conversation Analysis

Script to analyze costs:

import json
from pathlib import Path

def analyze_conversations():
    """Analyze all saved conversations."""
    
    conversations_dir = Path("conversations")
    json_files = list(conversations_dir.glob("*.json"))
    
    if not json_files:
        print("No conversations saved yet.")
        return
    
    total_cost = 0
    total_tokens = 0
    total_messages = 0
    
    print(f"\n📊 ANALYSIS OF {len(json_files)} CONVERSATIONS\n")
    print(f"{'ID':<20} {'Messages':<10} {'Tokens':<10} {'Cost':<12}")
    print("="*52)
    
    for file in json_files:
        with open(file) as f:
            data = json.load(f)
        
        print(f"{data['conversation_id']:<20} {data['message_count']:<10} "
              f"{data['total_tokens']:<10} ${data['total_cost']:<11.6f}")
        
        total_cost += data['total_cost']
        total_tokens += data['total_tokens']
        total_messages += data['message_count']
    
    print("="*52)
    print(f"{'TOTAL':<20} {total_messages:<10} {total_tokens:<10} ${total_cost:<11.6f}")
    print(f"\nAverage per conversation: ${total_cost/len(json_files):.6f}\n")

if __name__ == "__main__":
    analyze_conversations()

Run it:

python analyze_conversations.py

✅ Self-Assessment Rubric

Functionality (40 points):

(10 pts) The bot answers coherently
(10 pts) Maintains context (references to previous messages)
(10 pts) Special commands work (historial, costo, salir)
(10 pts) Conversations are saved to JSON

Error Handling (30 points):

(10 pts) Handles rate limit with retry
(10 pts) Handles API errors with retry
(10 pts) Error logging to a file

Optimization (20 points):

(10 pts) Optimized temperature (0.3 for support)
(10 pts) Limited max tokens (150)

Cost Tracking (10 points):

(10 pts) Cost calculated correctly

Total: ___/100 points

Interpretation:

90-100: ✅ Excellent
70-89: ⚠️ Good
<70: ❌ Review

🎯 Extensions (Optional)

1. Persistent memory across sessions:

# On startup, load the previous conversation
def load_last_conversation(self):
    # Find the latest JSON
    # Load the messages
    pass

2. Streaming responses:

response = client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=self.messages,
    stream=True
)

for chunk in response:
    print(chunk.choices[0].delta.content, end="", flush=True)

3. Web UI with Streamlit:

pip install streamlit

import streamlit as st

st.title("🤖 TechApp Support")
user_input = st.text_input("You:")
if user_input:
    response = bot.chat(user_input)
    st.write(f"Bot: {response}")

📊 Summary

What you built:

Production-ready chatbot:
- Conversational context
- Custom system message
- Optimized parameters
Robust error handling:
- Retry with exponential backoff
- Logging to a file
- Graceful degradation
Cost tracking:
- Automatic calculation
- Conversation analysis
Persistence:
- JSON per conversation
- Later analysis

Skills mastered:

✅ Full OpenAI SDK
✅ Conversations with context
✅ Error handling in production
✅ Cost optimization
✅ Logging and debugging

Congratulations! You completed Module 2.

🔗 Additional resources

OpenAI Best Practices
Streamlit Docs - For a web UI
Prompt Engineering - Improve system messages

➡️ Next step

Next module: Module 3 - LM Studio (Local GUI)

You'll learn to run LLMs locally on your machine:

Zero operating cost
100% privacy
OpenAI-compatible API (reusable code!)

Time: 2-3 hours

Estimated time: 60-90 minutes
Project completed! 🎉

← PreviousError Handling and Retry Strategies Next →Module 3: LM Studio - Introduction