Build AI Text Pipelines Using LLM Nodes in DataWorks - DataWorks

Large language model (LLM) nodes let you process unstructured text data directly within DataWorks pipelines. Instead of writing custom algorithms, use natural language prompts to perform AI tasks such as:

Text summarization
Sentiment analysis
Content classification
Information extraction

You can add these AI capabilities to existing extract, transform, and load (ETL) pipelines for data like user reviews, product descriptions, and customer service logs.

Prerequisites

Deploy an LLM service in DataWorks before you create an LLM node. For instructions, see Deploy a model.

Important

Your choice of model and resource specifications directly affects the performance and response speed of the model service. The model service also incurs resource group fees.

Node settings

Setting	Description
Model Service	The LLM service you deployed in the prerequisites step.
Model Name	The model to use from the selected model service.
System Prompt	Defines the model's role, capabilities, and behavioral guidelines. Supports the `${param}` syntax to reference parameters.
User Prompt	The question or instruction sent to the model. DataWorks provides four built-in templates. Supports the `${param}` syntax to reference parameters.

Parameter syntax

Both System Prompt and User Prompt support the ${param} format to reference node or pipeline parameters. For example:

Please select items that match ${catalog}.

In this prompt, catalog is a node or pipeline parameter whose value is resolved at runtime.

Example: Pass data through an LLM node

This example builds a three-node pipeline that passes a value from an assignment node through an LLM node and outputs the result with a MaxCompute SQL node.

Pipeline structure

Step 1: Create a model service

Open the DataWorks LLM service console.
Create a model service based on Qwen3-1.7B.
For Resource Group, select the resource group attached to your current workspace.

Step 2: Create the pipeline

Go to Data Studio and create a pipeline with three nodes: an assignment node, an LLM node, and a MaxCompute SQL node.

Step 3: Configure the assignment node

Set the language mode to Shell in the toolbar in the lower-right corner. For more information, see Assignment node.
Enter the following code:
```
echo 'DataWorks';
```

Step 4: Configure the LLM node

Select the Model Service and Model Name you created in Step 1.

Set User Prompt to:

Write an introduction about ${title} with a word limit of ${length}.

In the configuration pane on the right, go to Run Configuration > Resource Group. Select the same resource group you chose when you created the model service.

In the configuration pane on the right, go to Scheduling > Scheduling Parameters. Add two parameters:

Parameter	Value
`title`	Output of the assignment node
`length`	`300` (static value)

To attach the output from an upstream node, click the icon to the right of the value input box.

LLM node scheduling parameters

Step 5: Configure the MaxCompute SQL node

Important

The MaxCompute SQL node requires a MaxCompute computing resource. If you don't have one, use a Shell node instead to display the output.

Enter the following code:
```
select '${content}';
```
In the configuration pane on the right, go to Run Configuration > Resource Group. Select the same resource group you chose when you created the model service.
In the configuration pane on the right, go to Scheduling > Scheduling Parameters. Add the content parameter and set its value to the output of the LLM node (llmtest).

To attach the output from an upstream node, click the icon to the right of the value input box.

Step 6: Run the pipeline

Return to the pipeline view and click Run at the top.
Enter the run parameters in the pop-up window.

Expected output

After the pipeline runs successfully, the MaxCompute SQL node returns a result similar to:

DataWorks is an enterprise data development and management platform from Alibaba Cloud. It supports data collection, cleansing, integration, scheduling, and visualization for large-scale data processing.
It provides a visual interface, connects to various data sources, and features powerful task scheduling and data quality monitoring.
DataWorks handles both real-time and batch processing, helping enterprises manage data as assets and improve efficiency.
Its unified process helps build reliable data pipelines for data governance and intelligent analysis.