AI Agent Studio
Summarize
Summary of AI Agent Studio
AI Agent Studio enables ServiceNow customers to create, manage, and test AI agents and agentic workflows that automate business processes and help achieve operational goals. It requires the installation of Now Assist AI agents to activate the agentic AI experience. The Studio’s main Overview page provides access to ready-made templates, recent agentic workflows and AI agents, and guidance resources including a journey checklist and introductory video.
Show less
Managing AI Agents and Agentic Workflows
The Create and Manage page in AI Agent Studio offers two tabs to organize agentic workflows and AI agents separately. Users can efficiently search, filter, create, duplicate, delete, and configure these components via Guided Setup. This centralized management simplifies maintaining and updating agentic AI solutions within the ServiceNow environment.
Agentic AI Activity and Version Control
The Activity page logs execution history for AI agents and workflows, supporting filtering by criteria such as version number. This feature helps track performance and troubleshoot issues. Customers can implement version control to manage multiple iterations of AI agents and workflows for controlled deployment and testing.
Testing AI Agents and Workflows
AI Agent Studio provides manual and automated testing capabilities to validate AI agent and workflow performance. Manual tests simulate single executions to verify expected behavior, while automated evaluations analyze multiple runs to detect patterns or issues. Security controls can also be tested. Note that live agent assistance via Now Assist panel is not supported in this testing module; live interactions require Now Assist in Virtual Agent.
Settings and Configuration
The Settings page allows enabling Now Assist Guardian, which adds advanced protections such as offensiveness detection, prompt injection prevention, and long-term memory management for AI agents. These configurations help maintain AI agent reliability, security, and compliance with organizational policies.
Analytics and Performance Monitoring
The AI Agent Analytics dashboard presents key metrics including the number of active agents and workflows, efficiency gains in time-to-resolution, and execution counts. This data empowers customers to measure the impact of their AI agents and optimize their workflows for better business outcomes.
Create, manage, or test AI agents and agentic workflows so that you can create self-executing workflows to help you achieve your business goals.
AI Agent Studio overview
With the AI Agent Studio application, you can create, manage, or test AI agents and agentic workflows all in one place. To enable the agentic AI experience, you must first install Now Assist AI agents. For more information, see Install Now Assist AI agents.
The Overview page has three sections where you can find the information that you must understand, begin, and continue developing AI agents and agentic workflows. When you first go to the AI Agent Studio, tour points are available to guide you through the experience.
In the Ready-made agentic workflows and AI agents section, you can find the templates and ready-made AI agents and agentic workflows that you can incorporate as-is or in custom workflows that you create. You can even explore more templates to find the ones that best fit your needs.
In the Recent agentic workflows and AI agents activity section, you can find the agentic workflows or AI agents that have been created or configured most recently. On both List views, you can create, duplicate, or delete agentic workflows or AI agents. The tab also includes a link to the AI Agents Dashboard where you can review the details about AI agent usage and performance.
The third section is a card to open a modal with a journey checklist that describes the steps to incorporate the AI agents into your workflows successfully and a video that provides an overview of the AI agents and workflows to get you started. Select View steps to open this section.
The following example shows the Overview page of AI Agent Studio.
Managing agentic workflows and AI agents
From the AI Agent Studio Create and manage page, you can create, duplicate, or manage the existing agentic workflows and AI agents. This page has two tabs, one for agentic workflows and one for AI agents. You can edit the columns of the List view to change what information is displayed. You can also search or filter the Lists to find the agentic workflows and AI agents that you're looking for quickly. By selecting the name of the agentic workflow or AI agent, you can open Guided Setup to configure or reconfigure the agentic workflow or AI agent.
The following example shows the AI Agent Studio create and manage page after several Now Assist applications are installed.
Agentic AI activity
The activity page contains execution logs for both agentic workflows and AI agents. The list allows you to filter based on various fields, including the version. For more information on creating multiple versions of the List of steps field, see Version control for AI agents and agentic workflows.
The following example shows several execution logs in AI Agent Studio.
Testing agentic AI
There are two types of manual tests you can do: AI agent or agentic workflow to test execution or Test access to test security controls. You can view your executed tests in the two tabs of the testing page. For manual execution tests, you can select the reply button to repeat the test. You can open the full details page of an automated test by selecting the test's name in the list in the Automated tab.
The following example shows the inputs for a Generate Resolution Plan agentic workflow testing setup of AI Agent Studio.
AI Agent Studio settings
The following image shows the AI Agent Studio settings.
AI Agent Analytics dashboard
From the AI Agent Analytics dashboard, you can review the vital statistics about your AI agents and agentic workflows. You can see the data about the number of agents and agentic workflows used, time-to-resolution efficiency gain, and the number of executions.
The following example shows the overview page for the AI Agent Analytics dashboard.