With the spotlight on ServiceNow AI Agents during Knowledge 2025, interest in this innovative capability has surged. However, a common challenge many are facing is the inability to experiment with AI Agents on Personal Developer Instances (PDIs). This limitation is preventing users from applying the insights and skills they've gained during the event, creating a gap between learning and hands-on implementation.
In this blog series, we’ll explore ways to work around this limitation by leveraging ServiceNow learning instances to gain hands-on experience with AI Agents.
Steps to Obtain a ServiceNow Instance with AI Agent Studio
1. To obtain ServiceNow AI Agent enabled instance, login to https://learning.service-now.com/ and search for
Now Assist Skill Kit Essentials course or directly navigate to Now Assist Skill Kit Essentials and request for a lab instance. Please note that it might take around ~10 minutes for the instance to be up and running.
2. Once the instance is UP, navigate to AI Agent Studio -> Create and Manage to checkout AI Agent Studio. The AI Agent Studio will open with some OOB Use Cases and AI Agents.
3. Now, we will Turn On the Now Assist Panel but prior to that, we need to allocate some roles to our logged in system administrator user. Open the user record and assign the below roles to admin user:
- sn_aia.admin : Administrator of the Now Assist AI Agents application.
- now_assist_panel_user : Allows users to utilize the Now Assist Panel features
- sn_nowassist_admin.nsa_admin : This user can create and update the Now Assist Admin experience by editing and configuring skills.
Logout and login the admin user for the roles to take effect.
4. To turn on the Now Assist Panel, navigate to Now Assist Admin -> Experiences. Under the Now Assist Panel tab, click on Turn on button. This will enable the Now Assist Panel. Observe a small Now Assist Panel icon just adjacent to scope picker.
5. Next, we will test one of the AI Agent available OOB which is Knowledge Content Generator which could be used to generate knowledge articles from a record. Navigate back to Create and Manage under AI Agent Studio. Under the AI Agents tab, lookout for Knowledge Content Generator AI Agent. Have a look on the Agent description, role and the instructions provided to the agent. Once done, click on continue to see the list of tools being used by the AI Agents.
Once done, click on Continue and ensure that the Agent Status is set to Active and Running. Finally click, Save and test to test the AI Agent.
6. Under the Test Scenario tab, ensure AI agent "Knowledge Content Generator" is selected under what to test. Let's give a task to this AI Agent which involves generating a KB Article under a user selected Knowledge Base from an incident record. A sample prompt could be "Help me generate knowledge content from INC0010253". Once you are ready, click on Start test button to invoke the AI Agent.
The AI Agent execution will start and you can see the Output being generated under the Output tab. Also, the AI Agent decision logs could be reviewed on the right hand side. In the middle, you can view the Orchestrator invoking the AI Agent.
The AI Agent uses the tool "Fetch writable Knowledge bases" tool to ask user confirmation on the Knowledge Base under which the Knowledge Article needs to be created. Once the Knowledge Base is selected, the "Generate Knowledge Article content tool" generates the Knowledge Article under the selected Knowledge Base and provide the Article Number as an output.
Let's have a look on the Generated Knowledge Article. Notice that the KB Article Body generated is not as we expected. This is because of the fact that the Incident provided to generate the KB Article lacks sufficient knowledge thus, we can infer that the input data matters a lot while working with the AI Agents.
Congratulations! You have successfully explored your first AI Agent. In the next blog, we will build an Agentic AI Workflow (Use Case) with some AI Agents driving it.
If you find this blog helpful, please like and bookmark it for future reference.
