Conversational AI powers chatbots that exchange natural-language messages with users. A simple chatbot sends a prompt to an LLM; a business chatbot can also use private company data and tools. section 1.3.6 section 10.1.3
Customer service chatbot
A customer service chatbot can retrieve company information, help customers track or cancel orders, and escalate problems to a human agent. section 10.0
Listing 1.3 A working AI chatbot in less than 30 lines of code
import os
import streamlit as st
from openai import OpenAI
os.environ["OPENAI_API_KEY"] = "sk-..." # Replace with your own API key
openai = OpenAI()
human_message = lambda m: {"role": "user", "content": m}
ai_message = lambda m: {"role": "assistant", "content": m}
def talk_to_ai(question, history):
return openai.chat.completions.create(
model="gpt-5-mini",
messages=history + [human_message(question)],
).choices[0].message.content
st.session_state.history = st.session_state.get("history", [])
history = st.session_state.history
for message in history:
st.chat_message(message["role"]).markdown(message["content"])
prompt= st.chat_input("Chat with me!")
if prompt:
st.chat_message("human").markdown(prompt)
response = talk_to_ai(prompt, history)
history.extend([human_message(prompt), ai_message(response)])
st.chat_message("ai").markdown(response)
See a compact Streamlit and OpenAI chatbot example with chat messages and conversation history. section 1.3.6
