Data Products Builder Copilot

Alation Cloud Service Applies to Alation Cloud Service instances of Alation

The Data Products Builder Copilot provides a guided and flexible way to define, configure, and publish data products in your Data Products Marketplace. The Builder Copilot lets you create a structured YAML specification for your data product that follows Alation’s data product model and format requirements. You enter actions and describe updates in natural language, and the Builder Copilot generates the corresponding YAML sections. You can then review the generated content and apply it to the editor. It also preserves any manual edits that you make in the YAML.

The Builder Copilot feature is available in the data product editor window, and users who create and edit data products will automatically have access to it.

Data Product Structure

A data product follows a prescribed format. It consists of components, each represented as properties in the YAML definition. Some properties are required, while others are optional.

Required Properties

Every data product specification must include the following:

  • Contact email (contactEmail): Contact email of the data product owner

  • Contact name (contactName): Name of the data product owner

  • Product ID (productID): Unique product identifier

  • Version (version): Version of the data product

  • Name (name): Name of the data product

  • Description (description): Description of the data product

Optional Properties

You can also add optional components, such as:

  • Delivery systems (deliverySystems): Configuration that enables data consumers to request access to the data in the product

  • Record sets (recordSets): A list of data assets that make up the product, with descriptions and optional mock data

You can find more details about all properties in Get Started with Creating Data Products.

The Builder Copilot structures all YAML properties in the required format based on your natural language prompts. You guide the Builder Copilot by specifying:

  • Which component of the specification you want to add

  • What information to include in the relevant properties

Get Started with Builder Copilot

The left pane on the data product editor is your Builder Copilot. You can chat with it in natural language, and it writes or edits the YAML specification. If you’re satisfied with the response, you can Apply the Builder Copilot Output to the editor on the right. Use the Builder Copilot to fill in product metadata, delivery systems, record sets (tables) with fields, and example data. You can use the Preview feature (on the top left of the editor) to see a rendered view, and you still have full control to edit, validate, and save the YAML specification from the editor.

The screenshot below shows the Builder Copilot chat interface on the left of the data product editor window.

../_images/DPB_CopilotPane.png

Ready-To-Use Prompts

Data products Builder Copilot offers three ready-to-use prompts. Click a prompt to send it to the chat field at the bottom, then press Enter to send. The following sections explain the responses to the ready-to-use prompts.

What can a data product builder do?

Use this prompt to kickstart the Builder Copilot from scratch. Once you click this prompt and press Enter, Builder Copilot sends you back the following message, which describes the purpose of the Builder Copilot and the properties that you can ask Builder Copilot to edit in the YAML format.

Status: Beta
Description:
Data Product Builder is
your copilot. It helps you
define your data product
from scratch or refine an
existing product. You can instantly
preview and get feedback.
Ask your builder to:
- Add table <table_name>
- Add table 'customers' with start_date, country, and acq_date
- Add revenue column
- Add acquired_date column
- Suggest sample data
- Change description
- Update owner info
- Set logo url
- Search and add existing tables from the Alation Catalog using the @ symbol
- Update/Delete existing tables and columns from the data product (without using the @ symbol)

Write a description for the data product

Use this prompt when you want Builder Copilot to draft or improve the description of the data product. Once you click this prompt and press enter, Builder Copilot sends you back the following message with a Builder Copilot-generated description. You can still edit the description in the editor pane or prompt Builder Copilot if you need any changes to the suggested description.

Along with the description, Builder Copilot also shares the following properties with the placeholders. Users can update the placeholders with the original value.

  • contactEmail

  • contactName

  • name

  • productId

  • version

product:
 contactEmail: [email protected]
 contactName: Product Owner
 productId: product_id_example
 version: 0.0.0
 en:
     name: Product Name
     shortDescription: Comprehensive data product overview
     description: |
     # Comprehensive Data Product Overview

     This data product provides a centralized, reliable source of structured
     data designed to meet diverse analytical and operational needs across
     your organization. Built with scalability and data quality in mind, this
     product offers seamless access to critical business information through
     multiple delivery channels.
     <...>

Add a sample table with columns and mock data

Each data product lists the data available for consumers to use in the relevant use case. The data component is defined in the specification under the recordSets property.

Use this prompt to inspect a sample table with columns loaded with mock data. Once you click this prompt and press Enter, Builder Copilot responds with the following message, showing the structure of the recordSets component, which allows listing your data assets.

product:
    contactEmail: [email protected]
    contactName: Product Owner
    en:
        description: |
        # Product Description In Markdown
        This is a description of the product.
        name: Product Name
    productId: product_id_example
recordSets:
    sampleTable:
    description: &gt;
        A sample table containing customer information with basic demographic
        and contact details.
    displayName: Sample Table
    name: sample_table
    sample:
        data: |-
        customer_id,first_name,last_name,email,registration_date,is_active
        1,John,Doe,[email protected],2023-01-15,true
        2,Jane,Smith,[email protected],2023-02-20,true
        3,Bob,Johnson,[email protected],2023-03-10,false
        4,Alice,Brown,[email protected],2023-04-05,true
        5,Charlie,Wilson,[email protected],2023-05-12,true
        type: mock
    schema:
        - description: Unique identifier for each customer
        displayName: Customer ID
        name: customer_id
        type: integer
        - description: Customer's first name
        displayName: First Name
        name: first_name
        type: string
        - description: Customer's last name
        displayName: Last Name
        name: last_name
        type: string
        - description: Customer's email address
        displayName: Email Address
        name: email
        type: string
        - description: Date when customer registered
        displayName: Registration Date
        name: registration_date
        type: date
        - description: Whether the customer account is active
        displayName: Active Status
        name: is_active
        type: boolean
version: 0.0.0

You can use Free-Form Prompting with Copilot to refine the results until they match your requirements.

List Data in Your Data Product

Use targeted prompts to add tables and columns:

  • Add table <table_name>

  • Add column <column_name>

Examples:

  • Add table customers with start_date, country, and acq_date

  • Add the revenue column

  • Add the acquired_date column

For tables already present in the Alation catalog, you can use at-mention search to find and reference them:

  1. In the chat input field, type @.

  2. Search for and select the table you want to add.

../_images/DPB_TableSearch.png

When you send a prompt, Builder Copilot responds with generated YAML for the recordSets property.

Review the output and further refine it by:

  • Adding or removing tables and columns

  • Asking Builder Copilot to generate mock data

Free-Form Prompting with Copilot

The three ready-to-use prompts are just shortcuts. You can type any instruction in plain English, and Builder Copilot will add, edit, or refactor the YAML while preserving what’s already there.

The following are a few prompt examples that you can refer to when creating your data product:

  • Create a data product product named Transaction Details, product ID cust_trans_information, owner Admin.Account@Alation.com, version 1.0.0.

  • Add tables that contain transaction information, including table descriptions and display name.

    Note

    If you know which table you want to be added in the data product, you can type @ followed by the exact name and select it.

  • Suggest realistic sample data for the tables.

  • Improve short and long descriptions for business and technical audiences.

Prompt Tips

  • Use one intent per message to keep results precise.

  • Be explicit: Name the record sets and list fields with types (string, number, integer, Boolean, date, timestamp).

  • Use IDs vs. display names: Name for stable IDs (cnf_ordrs), displayName for human-readable labels (Confirmed Orders).

  • Iterate: Use Builder Copilot to make changes, then refine to get to the result you want.

Apply the Builder Copilot Output

Every Builder Copilot reply includes a small toolbar to help you move YAML between the chat and the editor.

../_images/DPB_ResponseToolBar.png

Copy YAML

The Copy YAML action is represented by the clipboard icon (region 1 on the screenshot above). It copies the entire YAML from that Builder Copilot message to the clipboard, so you can paste it into the editor. Manually copy the YAML, If you only want a portion of the YAML to editor.

Update YAML

The Update YAML action is represented by the arrow icon (region 2 on the screenshot above). It inserts the YAML from the response directly into the editor. This replaces the corresponding section of the specification. If the response includes a whole YAML definition, the whole definition will be replaced.

Revert YAML to Previous Version

The Revert YAML action is represented by the counter-arrow icon (region 3 on the screenshot above). Restores the editor to its previous state before the last Apply from the toolbar. Use this if an applied change wasn’t what you expected.

Frequently Asked Questions

The Data Products Builder Copilot is an AI-powered assistant integrated into the data product editor. It helps users create, edit, and refine data product specifications in YAML format using natural language prompts.

  • Data product builders such as data engineers, analysts, and stewards who design and accelerate creation of governed data products.

  • Domain experts who provide business context and requirements, leaving technical definition to builders.

  • AI developers who require governed, AI-ready products as inputs into their workflows.

Users can fully review, edit, or override any YAML suggestions.