Use Poolside models in n8n

​

Watch the setup video

​

Prerequisites

• You run a self-hosted n8n instance. These steps use n8n’s unverified community node installation flow. For availability and permissions, see n8n community node installation.
• Your environment meets the Poolside node compatibility requirements.
• You have one of the following:
  • A Poolside Platform API key. To create one, go to platform.poolside.ai, open API Keys, and click New key.
  • The OpenAI-compatible base URL and API key, if required, for your self-managed model endpoint.

​

Steps

1. Install the @poolsideai/n8n-nodes-poolside community node.
   • From n8n
   • Manual installation

   This method requires an Owner or Admin role in n8n.

   1. In n8n, navigate to Settings > Community Nodes.
   2. Click Install.
   3. Enter @poolsideai/n8n-nodes-poolside in Enter npm package name.
   4. Select I understand the risks of installing unverified code from a public source.
   5. Click Install.

   With shell access to the n8n host or container, install the package in n8n’s default community nodes directory:

   mkdir -p ~/.n8n/nodes
   cd ~/.n8n/nodes
   npm install @poolsideai/n8n-nodes-poolside
   

   Restart n8n after installation.
2. Add an AI Agent or another chain node that accepts a language model to your workflow.
3. Click the Model connector below the node, then select Poolside Chat Model.
4. Create a Poolside API credential and configure it:

   Field	Value
   API Key	Your Poolside Platform key or the key configured for your self-managed model endpoint. If the endpoint does not validate API keys, enter any non-empty value, such as local-test.
   Base URL	Keep the default https://inference.poolside.ai/v1 for Poolside-hosted inference.

   If you run Poolside model inference on your own infrastructure, use the model’s OpenAI-compatible endpoint at https://<model-hostname>/v1. Use the API key configured for the endpoint, if it requires one.
   Save the credential. A green check confirms that n8n can reach the base URL and access its models endpoint.
5. Select a Poolside model from the Model dropdown. The node loads this list from your configured endpoint.
6. Optional: Under Options, configure Sampling Temperature, Maximum Number of Tokens, Timeout, Max Retries, or Thinking.
   Use Thinking for open-ended reasoning, code analysis, and generation. Turn it off for Text Classifier, Information Extractor, Sentiment Analysis, or workflows that use a Structured Output Parser. Reasoning text can prevent these nodes from parsing the required format.
7. Connect a trigger and any output nodes your workflow needs, then run the workflow. The setup works when the AI Agent or chain returns a response from the selected Poolside model.

​

Workflow examples

​

Triage support tickets

1. Trigger the workflow when a help desk, email, or webhook node receives a ticket.
2. Prompt the Poolside model to classify the ticket by category and urgency.
3. Route the result with a Switch node to the appropriate queue, alert, or system.

​

Extract structured data

1. Trigger the workflow on a schedule or when n8n receives unstructured text.
2. Connect the Poolside Chat Model to a node such as Information Extractor or to a chain that uses a Structured Output Parser.
3. Turn off Thinking, then define the fields the model must return.
4. Map the structured output into a database, spreadsheet, or another destination.

​

Customize a workflow

• Change the trigger and destination: Connect any n8n integration that fits your source and output systems.
• Change the prompt: Ask the model to classify, summarize, extract, moderate, or generate content.
• Change the model: Select another model available from your configured Poolside endpoint.
• Change model options: Adjust temperature, token limits, timeouts, retries, or thinking behavior for the task.

​

Related resources

• Poolside node for n8n source code
• Poolside node for n8n on npm
• n8n community nodes
• n8n workflow templates
• Poolside API
• Supported models