Review the performance of an AI worker in AI Agent Studio

  • Versão de lançamento: Australia
  • Atualizado 12 de mar. de 2026
  • 2 min. de leitura
  • Review the performance analytics of an AI worker to track their task execution success. You can use the analytics to help make choices to tune the AI worker to suit your exact business needs.

    Antes de Iniciar

    Role required: admin

    Por Que e Quando Desempenhar Esta Tarefa

    Monitoring AI worker performance allows you to identify places where your AI worker is succeeding or failing. With the information provided in the analytics, you can make choices about whether to change any aspects of the AI worker's profile or tasks to enable it to perform better.

    Frequently monitoring performance enables you to identify areas of concern earlier.

    Procedimento

    1. Navigate to All > AI Agent Studio > Create and manage.
    2. In the AI workers tab, select the AI worker you want to view the performance of.
    3. Navigate to the Performance tab in the AI worker guided setup.
    4. Review the performance analytics of the AI worker.

      The following are the different tabs and performance visualizations for AI workers that use the ZTSD Worker Template for the IT Service Management AI agent collection. You can get the agent collection by installing Now Assist for IT Service Management (ITSM).

      Effectiveness

      Provides the graphical representation in percentages to measure the effectiveness of the AI worker in resolving incident from various categories. You can view incident outcomes, accuracy, and interaction effort by category to understand the areas where the AI worker performs effectively or needs improvement.

      • Incident outcome by category: Graphical representation of the outcome of the incidents in percentages such as resolved, reassigned, unresolved, reopened or new based on the available category. The graph illustrates how effectively incidents from different categories are managed by the AI worker, resulting in positive outcomes.
      • Average exchanges for resolved tickets by category: Graphical representation of the incidents in percentages where average information exchanges take place between the AI worker and user based on various category. This graph illustrates how effectively the AI worker interacted with the user to resolve the incident.
      Efficiency

      Details resolution time, track speed, and latency to ensure the AI worker meets service level agreements (SLAs).

      • Mean time to resolve: Displays the average of how much time the AI worker takes to resolve the incident.
      • First response time: Displays the percentage number of incidents that the AI worker resolved or reassigned.
      Value and feedback

      Measure the overall value added, assess the impact, usage and user satisfaction.

      • Aggregated sentiment analysis: Shows the overall tone (from negative to positive) of user messages in incidents handled by the AI worker.
      • Inferred CSAT: Estimates customer satisfaction based on resolution success rate, number of follow-up questions, escalation frequency, and sentiment in user messages.
      • Adoption coverage: Shows how many relevant assignment groups have adopted and used the AI worker to handle incidents.
      • Direct user feedback: Graphical representation of the direct user feedback provided by customers in response to the incident resolved by the AI worker.

    O que Fazer Depois

    If you want to make changes to your AI worker based on the performance analytics, see Edit the profile of an AI worker or Edit the tasks of an AI worker.