What's new in Sigma

AI

Ask Sigma is now Sigma Assistant

Ask Sigma is now Sigma Assistant, and includes the following:

• Improved accuracy and performance when answering questions and performing analysis.
• Higher quality data source selection, relying on a new semantic search service that indexes configured sources using the AI provider set up for your organization.
• Improved experience, including the ability to provide feedback on the quality of responses.
• Highlighted sources are now configured sources, and the steps to add configured data sources for Assistant are updated. Configure sources for Sigma Assistant to answer questions without selecting a data source.
• Configure data models as data sources to improve accuracy and capabilities. When Sigma Assistant uses data models to answer questions, it can choose which relationships to use.

If you embed Ask Sigma, Sigma Assistant is accessible from the same URL and existing embeds continue to work.

For more details, see Ask natural language queries with Sigma Assistant.

Add AI context to data models (Beta)

You can now add AI context to a data model to help Sigma Assistant better understand how to use your data model as a data source. Adding AI context to a data model can help improve the accuracy and consistency of Sigma Assistant responses when answering questions that use the data model as a source.

For more details, see Manage AI context for data models.

Sigma MCP Server available

Use natural language to interact programmatically with your Sigma organization using the Sigma MCP server. You can now search, describe, and query data within your AI assistant's interface. You can connect to the Sigma MCP server from any AI assistant that supports connection to custom remote MCP servers via HTTP and OAuth.

For more details on supported tools, see Use the Sigma MCP Server.

AI apps

Custom sort action (GA)

The Custom sort action is now generally available.

This action enables you to create an action that sorts one or more columns in a target element.

For more information, see Create actions that modify or refresh elements

API

Create and manage data models as code (GA)

You can create and manage data models programmatically using the Sigma API. The endpoints use a code representation of the data model to retrieve contents, make updates, and create new data models. For more information, see Manage data models as code.

The following endpoints are generally available to read, create, and update data models programmatically:

• Get the code representation of a data model (GET /v2/dataModels/{dataModelId}/spec)
• Create a data model from a code representation (POST /v2/dataModels/spec)
• Update a data model from a code representation (PUT /v2/dataModels/{dataModelId}/spec)

For example representations of sources and elements, see the Data model representation example library.

To provide context to AI agents building data models from code, see Access Sigma documentation from AI tools.

New API endpoint to list scheduled exports for a user

The following endpoint to list all scheduled workbook and report exports owned by a specific user is now available:

• List scheduled exports for a user (GET /v2/members/{memberId}/schedules)

New query parameter for List data models endpoint

The List data models (GET v2/dataModels) endpoint now includes the skipPermissionCheck query parameter. When set to true, this parameter allows the API client to return all data models in a Sigma organization, including those not shared with or owned by the requesting user. The API client must have admin privileges to use this parameter.

Workbook elements

Control pan and zoom for maps

You can now choose whether to allow pan and zoom for maps. Select the Allow pan and zoom checkbox under Format > Map style to allow users to freely pan and zoom, or clear the checkbox to lock a map chart to a fixed position and zoom level.

Bug fixes

• Dataset migration now updates workbook references to point to the new data model in cases where the dataset was used as part of a join or union.

• If you edit a document using the API, such as by swapping a data source or editing a data model using the Update a data model from a code representation endpoint, there is now a prompt to discard stale changes and include the updates in the document the next time it is opened in the UI.

• Editing SQL for datasets now completes as expected and no longer runs infinitely.