Add a Dataverse knowledge source

Integrating Dataverse tables as your knowledge source allows you to ground your agent in the data contained in your tables. This process involves adding synonyms and glossary definitions of the tables and columns in your data. For more information, see Improve agent responses from Microsoft Dataverse.

To add Dataverse tables as a knowledge source, perform the following steps:

1. Open the agent.

2. Select Add knowledge from either the Overview or Knowledge pages.

3. Select Dataverse (preview).

4. Locate one or more of your Dataverse tables to add. Up to 15 Dataverse tables can be added per knowledge source. To narrow your selections, use the search field.

   Note

   Table recommendations are based on the name of your agent.

5. Preview the tables to ensure the appropriate tables were added. The preview only displays 20 rows and a set of columns, however, all the rows and columns are included in the knowledge source.

6. Review the knowledge name and description. The description should be as detailed as possible, especially if generative AI is enabled, as the description aids AI orchestration.

7. Optionally, to help improve the quality of the answers, add synonyms and glossary terms:

   1. Add synonyms for table columns that you selected. Select the Back button to accept changes.

   2. Add glossary terms to define domain-specific terminology and acronyms. Select the Back button to accept changes.

8. Select Add to finish adding the knowledge source.

Synonyms and Glossary terms

Synonyms, glossary terms, and definitions for the synonyms and glossary entries aid in AI orchestration. They provide grounding data to improve generated responses. By providing extra information for the AI to understand and interpret the information in the tables, you increase the likelihood of the AI to recognize your users requests, and return responses based on the information provided to the AI.

For scenarios where your Dataverse table contains a column composed of numeric values, you need to provide a synonym for the AI to understand what's in the column. For example, your agent is providing travel assistance, and the Dataverse table contains a column named "cr_123_abc" that uses flight numbers to correspond to cities.

Since the AI doesn't know how to qualify this information, it must be explicitly told how to interpret it. So, the maker adds a description for this column, such as the following: "cr_123_abc represents the departure city for each flight represented by the flight code."

Sample glossary definitions

Glossary definitions are used to paraphrase the terminology in your Dataverse table, so your agent better understands user questions and can respond better.

The following table illustrates scenarios where adding definitions for glossary terms provides useful context for your agent.

Scenario	Glossary term	Sample description
Acronym	VP	"VP" refers to the Vice President value in the "JobTitle" column of the "Contact" table.
Custom ownership	activity owner	The "activity owner" is identified by the "PartyId" column in the "ActivityParty" table.
Custom field	opportunity revenue	"Opportunity revenue" refers to the "Custom Revenue" column in the "Opportunity" table.
Complex rules or filter	overdue task	"Overdue task" refers to the the "task" table, when the "state code" column has an open value, and the "scheduled end date" column has a value that is earlier than today.

Note

• The descriptions in the table are examples. Test your descriptions to verify what descriptions return the best results.
• It might take up to 15 minutes for updated glossary terms and definitions to become available.