Edit

Share via

FAQ for Copilot in desktop flow activity (preview)

  • Article

Copilot in desktop flow activity enables administrators, CoE teams, business users, and makers to easily retrieve information about past flow runs by asking questions in natural language. For example, users can ask about the number of flows that ran yesterday, which machine ran the most flows, or how many flows are currently queued. In response, Copilot generates outputs that provide insights and answers to the questions asked.

What can Copilot in desktop flow activity do?

Copilot in desktop flow activity is capable of translating natural language queries into Microsoft Dataverse FetchXML query syntax, specifically optimized for desktop flow activity such as desktop flow runs, machines, and errors. This allows users to easily retrieve information about their automation data by asking questions in natural language. Additionally, Copilot determines the most suitable output visualization, such as a table, pie chart, bar chart, or line chart, to effectively present the insights and information to the user.

What is Copilot in desktop flow activity’s intended use?

The purpose of this system is to enable users to retrieve information about their automation activity data by asking questions in natural language.

How was Copilot in desktop flow activity evaluated? What metrics are used to measure performance?

The performance of Copilot in desktop flow activity was evaluated using a comprehensive set of metrics. The Copilot interface includes thumb up and thumb down gestures, allowing users to submit feedback on the AI outputs. This feedback is closely monitored to ensure that Copilot remains compliant, appropriate, and free of bias.

What are the limitations of Copilot in desktop flow activity? How can users minimize the impact of Copilot in desktop flow activity’s limitations when using the system?

One of the limitations of Copilot in desktop flow activity is that it currently only supports answering questions about desktop flow activity, and not yet about cloud flows activity. Additionally, the language currently supported with Copilot is English and only available in Dataverse environments provisioned in the United States region. To minimize the impact of these limitations, users can ensure that their queries are specific to desktop flow activity and are phrased in English.

What operational factors and settings allow for effective and responsible use of Copilot in desktop flow activity?

Effective and responsible use of Copilot in desktop flow activity can be achieved by ensuring that the user has the appropriate rights to access the desktop flow activity data in Dataverse. This means that Copilot only answers questions based on the data the user has permission to access. Additionally, it's important for users to understand the scope and limitations of Copilot, and to phrase their queries accordingly.

See also