Build an iMessage AI assistant with Python

Many ideas work better inside iMessage: you get replies, not bounces. In this tutorial you will send your first message, listen for inbound texts, and wire an AI reply using the SDK.

Prerequisites

• macOS Sonoma or Sequoia
• Python 3.11+
• Messages enabled and signed into your iCloud account
• Full Disk Access and Automation permissions for your terminal or IDE

Send your first message

from imessage_sdk import IMessageSender

sender = IMessageSender()
sender.send_text(phone_number="+15551234567", text="Hello from Python!")

This uses AppleScript under the hood so you do not need any private API. SIP stays on.

Stream inbound messages

Use the client to listen for new events and route them to your logic.

from imessage_sdk import IMessageClient
from imessage_sdk.models import IMessage

def on_message(msg: IMessage):
    if msg.is_from_me or not msg.content:
        return
    print(f"{msg.sender_phone or msg.chat_guid}: {msg.content}")

client = IMessageClient()
client.start_listening(callback=on_message)

Add an AI reply

from imessage_sdk import IMessageClient, IMessageSender
from imessage_sdk.models import IMessage
from openai import OpenAI

llm = OpenAI()
send = IMessageSender()

def on_message(msg: IMessage):
    if msg.is_from_me or not msg.content:
        return
    reply = llm.responses.create(
        model="gpt-5",
        input=f"Reply helpfully to: {msg.content}"
    ).output_text
    if msg.chat_guid:
        send.send_text(chat_guid=msg.chat_guid, text=reply)
    elif msg.sender_phone:
        send.send_text(phone_number=msg.sender_phone, text=reply)

client = IMessageClient()
client.start_listening(callback=on_message)

Send images and files

from imessage_sdk import IMessageSender
sender = IMessageSender()

sender.send_file(
    phone_number="+15551234567",
    file_path="./samples/product.jpg",
    caption="New drop. Thoughts?",
)

Deploy on cloud Macs

You can run this locally or on Mac virtual machines such as AWS EC2 Mac and MacStadium. Sign into iCloud, enable Messages, grant permissions once, then run your service. SIP stays on.

Next steps

• Add a simple router to handle multiple flows
• Use group chat GUIDs to work inside teams
• Store transcripts in your database for analytics

When you are ready to ship, get the SDK and start building today.