LangChain iMessage agent

Building on the iMessage API? Build on iMessage with Flow: one agent for iMessage, WhatsApp and Telegram. New to it? Read the iMessage API guide.

TypeScript: stream the answer

Flow’s LangChain guide starts a run for each incoming message and delivers the streamed answer as chat bubbles:

// npm install @flow-engineer/messaging @langchain/openai
import { ChatOpenAI } from "@langchain/openai";
import { FlowMessaging, contentText } from "@flow-engineer/messaging";

const flow = new FlowMessaging();
const model = new ChatOpenAI({ model: "gpt-4.1-mini" });

for await (const event of flow.events.stream({ types: ["message.received"] })) {
  const stream = await model.stream([
    ["system", "You are the assistant for Asha's Bakery. Answer in short messages."],
    ["human", contentText(event.data.message.content)],
  ]);
  await event.conversation.reply(stream); // reads each chunk's content
}

Python: reply in the webhook

The Python example answers inside the webhook response, which is one round trip:

# pip install langchain-openai fastapi
from fastapi import FastAPI, Request
from langchain_openai import ChatOpenAI

model = ChatOpenAI(model="gpt-4.1-mini")
app = FastAPI()

@app.post("/flow")
async def flow_webhook(request: Request):
    event = await request.json()  # verify Flow-Signature first
    if event["type"] != "message.received":
        return {}
    text = event["data"]["message"]["content"].get("text", "")
    answer = await model.ainvoke([("system", "You are the assistant for Asha's Bakery."), ("human", text)])
    return {"reply": {"type": "text", "text": answer.content, "format": "markdown"}, "fallback": "auto"}

Verify the Flow-Signature header before you use the body. Send iMessage programmatically shows how.

How it fits

• Any streaming runnable. Chat models, prompt.pipe(model) chains and chains ending in a string output parser all work with reply().
• A 10 second budget in Python. Flow waits 10 seconds for the webhook answer. That suits fast chains. For slower ones, answer {} at once and send the reply later with POST /v1/conversations/{conversation_id}/messages.
• Plain text where needed. fallback: "auto" sends markdown as plain text on channels without formatting. iMessage is one of them.
• One message, not bubbles. A reply in the webhook response goes out as a single message, so keep it short.

Memory and limits

Keep chat history per conversation ID, or read it from GET /v1/conversations/{conversation_id}/messages. Flow keeps message content for 30 days by default, so persist anything you need for longer.

The text limits are 9,999 characters on iMessage and 4,096 on Telegram. Use the event ID as the idempotency key when you reply, so a retried delivery cannot send twice.

Flow’s iMessage numbers are US numbers, shared or dedicated. Build and test on the Telegram sandbox with the same code, then move to iMessage. For how the whole agent should be designed, see Build an iMessage AI agent.

Other frameworks: Vercel AI SDK, OpenAI Agents SDK, Claude Agent SDK.

Frequently asked questions

Keep reading

All posts: the Flow blog. The Flow docs have the reference.