Configure Microsoft LUIS as the NLU provider for Virtual Agent
-
- UpdatedAug 3, 2023
- 3 minutes to read
- Vancouver
- Virtual Agent
The Vancouver release is no longer supported. As such, the product documentation and release notes are provided for informational purposes only, and will not be updated.
Use the intents, entities, and utterances defined in a Microsoft Language Understanding Intelligent Service (LUIS) application and apply them as an NLU model for your Virtual Agent conversations.
Before you begin
In Microsoft LUIS, do the following:
- Sign in to the LUIS portal and create an application to be used as an NLU model (requires a Microsoft Azure account.)
- Locate and copy these Microsoft LUIS keys:
- Authoring key that was automatically generated when you created your Microsoft LUIS account. The authoring key provides the authentication needed for your LUIS applications and for creating, training, and publishing them. You can find the authoring key by signing into LUIS, selecting your user account, and opening Account Settings.
- Prediction endpoint runtime key that you assigned to the resource
for your Microsoft LUIS application. Virtual Agent accesses the LUIS runtime query prediction
endpoint through this key. Note: If you are using more than one LUIS application (NLU model), you must provide the prediction endpoint runtime key for each application that you create.
In your ServiceNow instance, do the following:
- Make sure that the Glide Virtual Agent plugin is activated, as it installs the Proxy Agent to the Microsoft LUIS Natural Language Understanding server plugin (com.glide.nlu.msluis.intent.discovery) needed for this integration. If you upgraded from a previous release, the upgrade process automatically retains the LUIS keys that you provided.
- Role required: admin
About this task
You can set only one NLU service provider for your instance.
Procedure