Metis Assisted Create

Metis Assisted Create helps you create a Data Product by describing your requirements in natural language. Metis analyzes your requirements, recommends relevant data sources, creates an OpenSpec specification, generates the Data Product, and deploys it.

Instead of manually setting up the project and identifying the required data sources, you can provide your requirements and let Metis guide you through the creation process.

Metis Assisted Create Overview

Use Metis Assisted Create when you want to create a new Data Product with guidance from Metis.

It is useful when you want to:

• Describe your Data Product requirements in natural language.
• Get recommendations for relevant data sources.
• Generate and review the Data Product specification.
• Generate and deploy the Data Product with Metis.

Create Data Product with Metis

The Metis Assisted Create workflow guides you through seven stages.

Step	Stage	Purpose
1	Intent	Describe the Data Product you want to create.
2	Sources	Connect to Snowflake and select the data sources that will be used.
3	Planning	Metis analyzes the requirement and creates the Data Product specification.
4	Review Spec	Review and validate the generated OpenSpec specification.
5	Generation	Metis generates the Data Product based on the approved specification and selected sources.
6	Review Product	Review the generated Data Product configuration and output ports.
7	Deploy	Deploy the Data Product and monitor the pipeline until publication is complete.

Step 1: Describe Your Data Product

The Intent step is where you describe what you want to create.

Enter your requirement in natural language. You can describe the business purpose, expected outputs, data to be used, calculations, or other requirements for the Data Product.

For example:

Customer Lifetime Value Predictor that combines customer data, demographics, and combined sales data to predict future customer spending patterns and identify high-value customers.

Metis analyzes the requirement and uses it as the starting point for the Data Product creation process.

Enhance the Requirement with Metis

Metis can enhance your requirement to make it more detailed and implementation-ready.

Select Enhance my Data Product requirements with Metis to generate an enhanced version. Review the original and enhanced requirements, then select Keep Original or Use Enhanced Version.

Configure Your Project

The Configure your project section is optional. You can customize where and how the Data Product is created.

You can configure:

• DataOps Group – Select the group where the Data Product will be created.
• Data Product Name – Enter a name or leave it empty to generate one automatically.
• Runner – Specify a runner tag or use the group's default runner.
• Template Project – Select a template project or use the organization default.
• Custom Skills – Select reusable organization-specific instructions for Metis to follow.

Custom Skills

Custom Skills let your organization define reusable instructions, such as naming conventions and modeling preferences, for Metis to follow when creating Data Products.

You can select existing skills or add a new skill with a name and instructions. Custom Skills are saved for your organization and can be reused when creating future Data Products.

After configuring the project, select Continue to proceed to Sources.

tip

Providing a clear business requirement and selecting the appropriate project configuration and Custom Skills helps Metis generate a Data Product that better matches your requirements and organizational conventions.

Step 2: Select Data Sources

After defining your requirements, continue to Sources.

In this step, Metis recommends relevant data sources based on your requirements. You can select from existing data sources or connect to a new Snowflake account.

Choose a Data Source

The Select Data Sources step provides two options:

• Existing Data Source – Select from data sources that are already available in your organization.
• New Data Source – Connect to a Snowflake account and select the data you want to use for the Data Product.

Metis initially displays data sources recommended based on your Data Product requirements.

Use an Existing Data Source

Select a recommended data source, or choose Browse All Data Sources to search and select another available data source from your organization.

Connect to a New Data Source

Select New Data Source to connect to a Snowflake account.

Choose an authentication method:

• Username-Password — Enter Account, Username, Password, Warehouse, and Role.
• Key-Pair — Enter the required key-pair connection details.

Select Connect to establish the connection.

note

Metis recommends using a Service User for the Snowflake connection. Make sure the account has the required permissions to access the data needed for your Data Product.

Select Databases, Schemas, and Tables

After successfully connecting to Snowflake, Metis displays the available databases.

Select the database you want to use. You can then browse its schemas and tables and select the data required for your Data Product.

Once you have selected the required data sources, select Continue to proceed to Planning.

tip

Choose only the databases, schemas, and tables required for your Data Product. This helps Metis generate a more relevant specification during the Planning step.

Step 3: AI Planning

During the Planning step, Metis creates an OpenSpec specification for the Data Product.

Metis prepares the workspace, analyzes the selected sources, and processes the requirement to determine how the Data Product should be implemented.

The planning process can include activities such as:

• Preparing the workspace and AI assistant.
• Analyzing the project context, OpenSpec artifacts, and source Data Products.
• Determining and updating the Data Product structure and specification.

Answer Questions from Metis

Metis may ask questions to gather additional information required to complete the specification.

Enter your response and submit it to continue the planning process.

tip

Review Metis's question carefully and provide the information required to complete your Data Product specification.

Step 4: Review the Specification

After planning is complete, Metis presents the generated OpenSpec specification for review.

The Review Specification step allows you to inspect the proposed implementation before Metis generates the Data Product.

The specification can contain requirements describing:

• Data transformations, calculations, and business logic.
• Output tables, columns, and data structures.
• Data Product behavior and refresh or materialization requirements.
• Expected scenarios and conditions.

Review the proposed changes carefully.

If the specification accurately represents your requirements, select Continue to proceed with generation.

note

Review the specification before continuing. This is the main checkpoint where you can validate that the proposed implementation matches the intended Data Product behavior.

Step 5: Generate the Data Product

After the specification is approved, Metis moves to the Generation step.

Metis generates the Data Product configuration based on:

• Your original or enhanced requirement.
• The selected data sources.
• The generated OpenSpec specification.
• Information collected during the planning process.

The generation process runs through a series of activities, including preparing an isolated workspace, starting the AI assistant, loading project context, and sending the request to the agent.

Metis continues processing until the Data Product has been generated.

Step 6: Review the Generated Data Product

Once generation is complete, Metis displays a Generation Summary.

The summary provides an overview of the generated Data Product and allows you to review the generated configuration before deployment.

Review the generated Data Product and confirm the configuration and output ports before continuing to deployment.

Step 7: Deploy the Data Product

The final step is Deploy.

Metis pushes the generated workspace and starts the pipeline responsible for compiling, validating, and publishing the Data Product.

The deployment screen provides visibility into the pipeline jobs and their current status.

Pipeline stages can include activities such as:

• Initializing the project, loading secrets, and setting up Snowflake.
• Verifying source access and populating the workspace.
• Building AI models, analyzing column lineage, and generating the Data Preview.
• Compiling and publishing the Data Product.

Each job displays its current status, allowing you to monitor deployment progress.

Publish the Data Product

After the required deployment jobs complete successfully, the Publish Data Product stage is completed.

You can then select Open in Operate to access the deployed Data Product and its operational information.

Creation Complete

When the deployment and publication process is complete, Metis displays a confirmation that the Data Product is ready.

The completion screen provides the deployment and review status and allows you to navigate to the Projects area.

A successfully completed workflow results in a Data Product that has been:

• Generated from the provided requirements.
• Configured using the selected data sources.
• Validated through the generated specification.
• Deployed through the Data Product pipeline.
• Published and ready for use.

note

Metis assists with Data Product creation, but generated requirements, specifications, and Data Product configurations should be reviewed before deployment to ensure they meet your organization's requirements and governance standards.