AI Assistant UI Guide

In this example, AI Assistant returns additional information regarding the existing XDM schemas in your sandbox, including their total count and the five most commonly used fields.

To view citations and validate AI Assistant’s response, select Show sources.

AI Assistant updates the interface and provides you with links to documentation that corroborate the initial response. Additionally, when citations are enabled, AI Assistant updates the response to include footnotes to indicate the specific parts of the answer that reference the provided documentation.

In the example below, AI Assistant is asked the following query: “Show me dataflows that were created using the Amazon S3 source”.

AI Assistant then responds with a table listing your dataflows and their corresponding IDs. Select the download icon ( ) to download the table as a CSV file. To view the entire table, select the expand icon ( ).

An expanded view of the table appears, providing you with a more comprehensive list of dataflows based on the parameters of your query.

When prompted with an operational insights question, AI Assistant provides an explanation of how it computed the answer. In the example below, AI Assistant outlines the steps it took in order to identify the dataflows that were created using the Amazon S3 source.

You can also provide filters and modifications to your questions, and you can instruct AI Assistant to render its findings based on the filters that you include. For example, you can ask AI Assistant to show you a trend of the count of segment definitions in the order of their created date, remove segment definitions with zero total profiles, and use month names instead of integers when displaying the data.

After receiving an answer for an operational insights question, select Show sources and then select View source query.

When queried with an operational insights question, AI Assistant provides an SQL query that you can use to verify the process that it took to compute its answer. This source query is for verification purposes only and is not supported on Query Service.

You can use autocomplete by including the plus symbol (+) in your query. As an alternative, you can also select the plus sign (+) located at the bottom of the text input box. A window appears with a list of recommended data objects from your sandbox.

In the example below, AI Assistant is first asked for the total number of dataflows and then is asked to list the 10 most recent dataflows.

To reset, select the ellipses (...) on the AI Assistant interface and then select Start new conversation. This informs AI Assistant that you intend on changing topics and can be particularly helpful when troubleshooting queries that are either failing or referencing incorrect information.

To view discoverability, select the lightbulb icon on the top header of the AI Assistant interface.

Next, select a category and then select a prompt from the list provided. You can use this feature to get a better idea of the types of questions that AI Assistant can answer. You can also update the pre-existing prompts with specific details that pertain to your sandbox using free text or autocomplete.

To view the panel of suggested questions, type at least seven (7) characters in the input box. Next, select the question that is relevant to you from the menu that appears.

You may need to update placeholders in some instances where a suggested question involves operational insights. For example, you may need to add the specific name of a dataset or an audience if the suggestion from AI Assistant includes placeholders.

Placeholders are highlighted in blue. Select the placeholder to start updating its value. For best results on numerical placeholders, ensure that you use digits instead of text. You can also use the entity autocomplete feature to update the placeholder values. You cannot a send a question that has unfilled placeholders.

NOTE: Suggestions are enabled by default. Select the Suggest ideas toggle to disable the feature.

Related suggestions are returned with every response from AI Assistant. To continue your conversation, select any of the suggestions in the related suggestions section.

Similar to placeholders in question autocomplete, you will need to update placeholders that are included in related suggestions before you can submit the query.