ASK : How to implement the async in langflow

[tabaolan1987]

Hi everyone,

I am new to Langflow. I am trying to build a chatbot that helps users generate customer information in a PDF file.

My intended flow is as follows:

User input:
“Please help generate the customer XXXX.”

Send the request as JSON to a server API using the API Request component.

The server immediately responds that the request is being processed.

The Chat Output component displays the message:
“The request is being processed.”

After the server finishes generating the PDF, it returns a download URL.

How can I display the download URL in the chat output after showing the “processing” message?

[KeepALifeUS]

Here's a pattern for async processing with intermediate messages:

State-Based Async Pattern

# State tracks the async operation
state = {
    "request_id": "req_123",
    "status": "processing",
    "processing_message": "Generating your PDF, please wait...",
    "result_url": None
}

# Step 1: Immediate response
def handle_request(user_input, state):
    # Start async operation
    request_id = start_pdf_generation(user_input)
    state["request_id"] = request_id
    state["status"] = "processing"
    
    # Return immediate message
    return state["processing_message"]

# Step 2: Polling/callback for completion
def check_completion(state):
    result = check_pdf_status(state["request_id"])
    if result.completed:
        state["status"] = "completed"
        state["result_url"] = result.url
        return f"Your PDF is ready: {result.url}"
    return None  # Still processing

For Langflow Specifically

Option 1: Webhook Pattern

1. Function node sends request to server, returns processing message
2. Server calls webhook when PDF is ready
3. Webhook triggers another flow that sends the URL to user

Option 2: Polling in Custom Component

class AsyncPDFComponent(Component):
    def build(self, request_id: str):
        # Poll for completion
        while True:
            result = check_status(request_id)
            if result.completed:
                return result.url
            time.sleep(5)  # Wait and retry

Option 3: Two-Phase Flow

1. First flow: Returns "Processing..." immediately
2. Store request_id in external state (Redis/DB)
3. Second flow (triggered later): Retrieves URL and sends to user

The key is decoupling the immediate response from the final result delivery.