Recommended actions provide context-aware suggestions to agents based on record details, enabling more efficient decision-making by presenting optimal field values or actions. This article focuses specifically on side panel-related recommendations.
Key Concepts in Recommendations
To effectively utilize recommendations, it's essential to understand the following components:
- Contexts
- Rules
- Recommendations
- Resource Generators
- Action Types (Guidances and Decision Trees for side panel recommendations)
As an example, we’ll create a recommendation that suggests incidents from which agents can "copy resolution notes." This recommendation is also available with Advanced RA for ITSM.
Contexts in Recommended Actions
A context (sn_nb_action_rule_context) determines which records from a specific table display recommendations when predefined rules are met.
If you have RA for ITSM installed, a default context record already exists for the incident table. For other tables, simply create a new context record for the desired table.
For this example, we will use the out-of-the-box (OOB) incident context.
Rules in Recommended Actions
Rules define the conditions under which recommendations appear. These conditions determine whether recommendations are displayed to agents with specific roles based on record attributes.
For our use case, we want recommendations to appear for active, non-child, and non-MI incidents, so we will use the following OOB rule:
Recommendations in Recommended Actions
Creating a Recommendation
Recommendations can only be created from an existing recommendation rule. To create a recommendation:
- Navigate to the rule form.
- Click New in the Recommendation related list.
- Select the appropriate recommendation type.
Field recommendations have been covered in a separate article, so we’ll focus on the Guidance and Decision Tree action types. Note that the Guided Decisions application (sn_gd_core) is required for the "Guided Flow to Follow" action type.
Guidance Action Type
"A specific action to take or information to review"
Agents can access recommended actions displayed as cards in the contextual side panel of the workspace. These actions can be executed directly by clicking the corresponding buttons, improving workflow efficiency.
Setting a guidance action type, will lead you to such a recommendation view on the SOW workspace:
For our example, we will create a Copy Resolution guidance, which is also available in the "Incident Management for SOW" plugin.
Creating a Guidance Action
- Create a guidance action in the ga_guidance_action table.
- For our use case we need two inputs, one for the source incident from which we need to copy the resolution and the other for the target incident, so create these two in guidance inputs:
We want the recommendation in the side panel to come up something like this: - Configure the preview experience to display recommendations in the side panel. For our use case, we configure it like:
Feel free to configure your own preview and detailed experiences. - Scroll down to Guidance Actions and click New. For our use case we will create a guidance action as follows:
- Here I am using the OOB guidance action automation “Copy Resolution”, you can create your custom automation based on what you want to achieve. The automation flow inputs are populated based on the inputs the guidance action automation takes. You can create a custom automation tailored to your needs.
Decision Tree Action Type
A Guided Flow is a structured, multi-step process that helps agents navigate decisions by answering a series of questions. The system then provides context-based guidance.
To enable this functionality, install the Guided Decisions (sn_gd_core) application.
You can refer here to explore on how to create these.
Resource Generators in Recommendations
Resource generators provide data inputs for actions such as guidances and field recommendations. There are various types of resource generators available. This is the documentation for different Resource Generator Types.
For this use case, we will use a similarity-based resource generator, which utilizes a task intelligence similarity definition to fetch the top N matching records.
I’ve the task intelligence model already trained, so I mapped the same to the resource generator. If you’re interested in a similarity or classification ML related solution that you want to use for recommendations, I suggest checking out Task intelligence in Servicenow.
Now we are good with creating the recommendation, on the rule of your choice, on the recommendation related list click new and select “A specific action to take or review” or for guidance decision trees, if you want to configure them.
Setting Up the Recommendation
To finalise your recommendation:
- Open the rule of your choice.
- In the Recommendation related list, click New.
- Select "A specific action to take or review" or configure a Decision Tree if applicable.
- Map the resource generator output to the guidance action inputs (created earlier).
- Save the configuration.
Observe, we have mapped the output coming from our resource generator to the guidance action inputs (the ones we created before).
Once set up, the recommendation will appear in the SOW workspace, enhancing agent efficiency.
Prebuilt ITSM Recommendations
Some ITSM-related plugins already provide built-in recommendations.
Troubleshooting: Why Aren’t Recommendations Appearing?
If you don’t see recommendations, check the following:
- Is the recommendation active?
- Does the record meet the rule’s conditions and role requirements?
- Is the resource generator input mapping correct?
- Is the guidance action input mapping properly configured?
- Have all relevant solutions been mapped correctly?
Questions?
Feel free to reach out with any questions. Hope this helps! 😊
Hi Saloni,
Thanks for the detailed explanation. I have configured it for CSM workspace but the Search bar and select source is not visible for some users, but for some users it is working properly. Proper roles are present for the users who cannot see this but still this is not visible, it shows 'No active recommendations'. I have added roles on Rules and added those role to the users as well but still the same issue.
Is there anything missing to configure?
Thanks in Advance
