Service request agent
The Service request agent understands what the user needs and automatically submits the correct service request, making service delivery faster and easier.
To enable and configure the Service request agent, do the following:
1.Log in to Neurons for ITSM as an Administrator.
2.Open the Configuration console and select AI Configuration Hub > Agentic AI > Agent Settings.
3.Enable the Service request agent using the toggle button.
4.Click Configure to further configure the agent.
5.Under Messaging and interaction, configure the following settings:
•Successful creation: Enter a success message that agent displays following the successful creation of a Service request. You can add {ServiceRequestId} and {ServiceRequestLink} to the message, which displays the service request ID and link to the record.
•Allow transfer to live agent: Enable the toggle button to transfer the interaction to your Service Desk Human Agent in case the Agentic AI bot cannot solve the user's request.
Along with this toggle, you must also enable the Allow transfer to live agent toggle in Global Settings > Escalation Handling.
6.Under Service request handling > Default service request template, set the default Service request template from the dropdown.
When the agent cannot find a specific Service request template for the user's query, it uses the default Service request template you've configured . Furthermore, when multiple Service request templates match the user’s query, the agent presents the template you configured as the default option to the user.
We recommend that you carefully design and thoroughly test all Service request templates before deployment, as the Service request agent might not always display templates as expected if they are not properly configured.
7.Under Advanced settings > Service request description, define the scope of the Service request agent by detailing the types of Service request queries the agent can handle. You can also provide some example queries related to Service requests to help the agent better understand the context. When users ask queries other than those defined here, the agent considers them out of context and replies to users accordingly.
Modifying the intent description might result in unpredictable agent behavior. We recommend that only users with experience in LLM prompt engineering make these changes.
The following image shows the configuration page of Service request agent.
Fields marked with an asterisk (*) are mandatory.
Limitations
•If multiple templates match a Service request prompt, only the top three are displayed. If none of these meets the user’s needs, the user should select the default Service request template.
•For some Service request templates, users may be prompted to enter the mandatory Requester field. The Adaptive Card dropdown list displays a limited number of entries. If the required requester is not visible within the list, the user cannot select the desired requester.
•Some Service request offerings generate errors upon submission of the Service request draft. The errors occur when the Service request draft have the following format:
Sample prompts that may generate this error include I need a new printer for my office and I would like to onboard a new vendor.
•During a Live Agent Transfer session, the End Conversation button is displayed only after the Live Analyst sends a response. Users can end the session only after the Live Agent has sent a response and the chat bot is waiting for input from the user. If the Live Agent does not respond, the session times out automatically, but the user cannot terminate the session.
•During Live Agent Transfer, a conversation summary should be sent to the Live Agent at the beginning of the session. This functionality is currently not working.
•At the end of a Live Agent Transfer session, users are not informed about what to do next. However, when the session ends, the Agentic AI is reset and users should initiate a new prompt, even if no live agent was available.
•The Agentic AI handles only IT‑related issues. Other Line of Business (LoB) queries, such as HR, Facilities, and other non‑IT requests are not supported.
•The Agentic AI does not fully support language translation. While it understands prompts in multiple languages, it currently responds using a combination of the user’s locale and English. The following image shows an example in which the user provides input in French, but the agent responds in English:
