AI Content Generation via Replicate API Polling - n8n Workflow

Automation specialists seeking production-ready n8n templates for complex, asynchronous AI tasks.
Developers needing to integrate the Replicate API's vcollos/trefilio model into larger data pipelines using n8n.
Users who require reliable polling logic within an n8n workflow to handle long-running AI generation jobs.
Anyone looking for a complete n8n solution for managing Replicate predictions, including success and failure routes.

Generating complex AI content often requires asynchronous processing, where an initial request starts a job, and the results must be periodically checked (polled) until completion. This powerful n8n workflow solves this complexity by providing a complete, resilient polling mechanism specifically tailored for the Replicate API.

Starting with a manual n8n trigger, the workflow securely authenticates, sets all necessary generation parameters (like prompt, size, and model specifics for the vcollos/trefilio model), initiates the request, and enters a smart loop. This loop uses Wait and httpRequest n8n nodes to check the status, ensuring the n8n workflow only proceeds once the generation is finished or failed. This structured approach makes this one of the most reliable n8n templates for AI content processing.

1. Trigger and Setup: The process begins with the Manual Trigger n8n trigger. Following this, the Set API Token n8n node stores the necessary Replicate API key, and the Set Other Parameters n8n node injects the generation inputs (like the prompt and image dimensions) into the workflow data.


2. Prediction Initiation: The Create Other Prediction n8n node sends a POST request to Replicate to initiate the job, receiving a prediction ID.


3. Initial Wait and Logging: The Log Request n8n node records the prediction ID for monitoring, followed by the Wait 5s n8n node, allowing the AI model time to start processing.


4. Status Polling Loop: The Check Status n8n node polls the Replicate API using the prediction ID to retrieve the current status.


5. Conditional Routing: The Is Complete? n8n node checks if the status is 'succeeded'.

1. Final Output: The resulting success or error object is processed by the Display Result n8n node, providing a clean, structured JSON output for the entire n8n workflow execution.

1. Import the n8n Workflow: Copy the provided JSON code and import it directly into your n8n instance via the Workflows section (New -> Import from JSON).


2. Acquire Credentials: Obtain your Replicate API token from their website (https://replicate.com).


3. Configure API Token: Locate the Set API Token n8n node and replace the placeholder value YOURREPLICATEAPI_TOKEN with your actual token.


4. Configure Parameters: Navigate to the Set Other Parameters n8n node. Here you can adjust the prompt, width, height, and other specific parameters required by the vcollos/trefilio model.


5. Execution: Ensure the workflow is active and click the 'Execute Workflow' or 'Test Workflow' button, starting with the Manual Trigger n8n trigger to generate your first output.

Manual Trigger (n8n trigger): The starting point of the n8n workflow, manually initiated to begin the AI content generation process.
Set API Token (n8n node): Essential setup n8n node that defines the apitoken variable used for secure authorization headers throughout the workflow.
Set Other Parameters (n8n node): Configures all input variables for the vcollos/trefilio model, including the core generation prompt, dimensions, and various optional settings like seed and guidancescale.
Create Other Prediction (httpRequest n8n node): Sends the initial POST request to https://api.replicate.com/v1/predictions. It dynamically constructs the JSON body using parameters from the previous n8n node and uses the API token in the 'Authorization' header.
Check Status (httpRequest n8n node): This n8n node is the core of the polling logic. It sends a GET request using the prediction ID to monitor the job's progress.
Is Complete? (If n8n node): A flow control n8n node that checks if the prediction status equals 'succeeded'.
Wait 10s (n8n node): Used in the retry loop. If the job is still processing, this n8n node introduces a delay before polling the status again, preventing API rate limit issues.
Success Response / Error Response (Set n8n node): These n8n nodes format the final output data, providing clear JSON objects indicating either the result URL or detailed error messages.