Building Your First AI Chatbot — A Python Tutorial for Beginners

You do not need a machine learning degree to build a working chatbot. This tutorial walks through two real, runnable versions: a simple rule-based chatbot using pure Python, and an upgraded version that calls an LLM API for open-ended responses. By the end you will have working code and understand exactly what each piece does.

Step 1: A Simple Rule-Based Chatbot

The simplest chatbot matches keywords in user input to predefined responses. No dependencies needed beyond core Python.

# rule_based_bot.py
def get_response(user_input):
    text = user_input.lower()

    if "hello" in text or "hi" in text:
        return "Hey there! How can I help you today?"
    elif "hours" in text or "timing" in text:
        return "We're open Monday to Saturday, 9 AM to 6 PM."
    elif "price" in text or "cost" in text:
        return "Our plans start at ₹499/month. Want details on a specific plan?"
    elif "bye" in text:
        return "Goodbye! Have a great day."
    else:
        return "Sorry, I didn't understand that. Could you rephrase?"

print("Bot: Hi! Type 'bye' to exit.")
while True:
    user_input = input("You: ")
    if "bye" in user_input.lower():
        print("Bot:", get_response(user_input))
        break
    print("Bot:", get_response(user_input))

This works for a narrow, predictable use case (FAQ bots) but breaks down quickly once users phrase things in ways you did not anticipate.

Step 2: Add Intent Matching with a Classifier

Instead of hardcoded `if/elif` checks, train a lightweight classifier on sample phrases per intent — this generalizes much better to varied phrasing.

# intent_bot.py
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.svm import SVC

training_data = [
    ("hi there", "greeting"), ("hello", "greeting"), ("good morning", "greeting"),
    ("what are your hours", "hours"), ("when are you open", "hours"),
    ("how much does it cost", "price"), ("what's the pricing", "price"),
    ("bye", "goodbye"), ("see you later", "goodbye"),
]

responses = {
    "greeting": "Hey there! How can I help you today?",
    "hours": "We're open Monday to Saturday, 9 AM to 6 PM.",
    "price": "Our plans start at ₹499/month.",
    "goodbye": "Goodbye! Have a great day.",
}

texts, labels = zip(*training_data)
vectorizer = TfidfVectorizer()
X = vectorizer.fit_transform(texts)

model = SVC(kernel="linear", probability=True)
model.fit(X, labels)

def get_response(user_input):
    X_input = vectorizer.transform([user_input])
    intent = model.predict(X_input)[0]
    return responses.get(intent, "Sorry, I didn't understand that.")

print(get_response("what time do you open"))  # -> hours response

Step 3: Upgrade to an LLM-Powered Chatbot

For genuinely open-ended conversation, call an LLM API instead of matching intents. This handles phrasing you never anticipated.

# llm_bot.py
import os
from anthropic import Anthropic

client = Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])

conversation_history = []

def get_response(user_input):
    conversation_history.append({"role": "user", "content": user_input})
    response = client.messages.create(
        model="claude-3-5-haiku-20241022",
        max_tokens=300,
        system="You are a helpful customer support assistant for a small SaaS company.",
        messages=conversation_history,
    )
    reply = response.content[0].text
    conversation_history.append({"role": "assistant", "content": reply})
    return reply

print("Bot: Hi! Ask me anything. Type 'bye' to exit.")
while True:
    user_input = input("You: ")
    if user_input.lower() == "bye":
        break
    print("Bot:", get_response(user_input))

Keeping `conversation_history` and passing it back on every call is what gives the bot memory of earlier turns in the conversation — without it, every message would be treated as a fresh, context-free question.

Choosing the Right Approach for Your Project

• Rule-based — best for a narrow, predictable FAQ bot with a handful of fixed intents and zero API cost
• Intent classifier — best when you have more varied phrasing but still a fixed, known set of possible intents
• LLM API-based — best when conversations are open-ended, or when you need the bot to handle questions you cannot fully anticipate in advance

Frequently Asked Questions