Skip to main content
About the Resource Center

Generate intents and utterances based on a description for Virtual Agent and Copilot

Virtual Agent uses large language models (LLMs) and generative artificial intelligence (AI) technology to create intents and utterances for you, which saves you the time and effort required to build and experiment with natural language understanding for the bot. Architect bot flow and digital bot flow authors create intents and descriptions, and Virtual Agent and Agent Copilot generate the utterances for them. When you generate intents and utterances, the more context that you provide, the better. For example, “The customer asks about a product’s availability in the online store.” Experienced flow authors may have a feel for how well the intent performs. Virtual Agent generates up to 10 intents and 20 utterances. You can adjust, delete, and add more utterances or intents for fine-tuning. You can also delete intents and utterances in bulk. Generating more utterances helps to improve natural language understanding. Architect automatically creates reusable tasks that you can configure for each intent.

If you have an existing intent that is similar to the one you just created, then the utterances that Virtual Agent generates can produce confusion with those other utterances in Intent Health. After Virtual Agent generates your utterances, you can edit, update, or delete your utterances. 

Notes:
  • To access the option to upgrade to Virtual Agent, you must purchase the Genesys Cloud AI Experience license.
  • After you publish and run the flow, Virtual Agent consumes Genesys Cloud AI Experience tokens. For more information, see Genesys Cloud AI Experience tokens metering and pricing.

Add Virtual Agent from a voice or digital bot flow

Upgrade each flow individually; you cannot upgrade to all voice or all digital bot flows at once. After you upgrade to Virtual Agent, generate intents, and complete the remaining configuration, publish the flow so that you can use the AI features.

For more information, see Add Virtual Agent to your voice or digital bot flow.

Descriptions in voice and digital bot flows

With Virtual Agent, the descriptions that you add to voice and digital bot flows in turn drive utterance generation. Virtual Agent uses the description that you add here and automatically populates the intent description in the Add Intent dialog box. The description appears at the top of the individual intent’s page.

Note: When you create intents in new voice or digital bot flows with Virtual Agent, you must include a description in the Add Intent dialog box. The Generate button remains unavailable until you add a description.

Genesys recommends that you create meaningful, detailed descriptions to provide context for utterance generation. For example, instead of “Customer wants to place an order,” more purposeful descriptions are, “Customer wants to place or modify a coffee order, which may include a specific brew, flavor, or size; customizations; and pick up or delivery options,” or “User wants to purchase a gift card and may ask about increment options.”

If the flow description does not align with the intent that you create, you can edit the intent description.

Generate intents and utterances with AI

Virtual Agent uses large language models (LLMs) and generative AI technology that enables Architect bot flow and digital bot flow authors to create or generate intents and descriptions. Virtual Agent and Agent Copilot also generate intents and utterances for AI-generated or user-created intents. 

To enable you to focus on creating and managing your intents and descriptions, in the Intents view, Show Health is turned off by default:

Also, when you click an intent and open the intent details, the intent’s slot assignments and utterances appear in expandable sections. AI-generated utterances are designated by the Virtual Agent symbol . User-generated utterances are not.

The Add Intent dialog box includes an Intent Description box, which drives the utterances that belong to the intent. Create a mandatory description, which enables the Generate button. The description generates the utterances.

  • If you enable Virtual Agent when you create the bot flow and then generate intents, Architect populates the Generate Intents dialog box with the bot flow description. You can modify or create a new description.
  • If you edit a Virtual Agent-generated utterance, including slot mapping or utterance wording changes, the utterance automatically becomes a user-generated utterance and is then preserved from future regeneration. Even if you revert to the original utterance wording, the utterance remains a user-generated utterance.
Note: You can have 20 utterances. If you have no utterances, Virtual Agent generates 20 utterances. If you already created five utterances, Virtual Agent only generates 15 utterances. If you have 20 utterances, Virtual Agent does not create any utterances. User-created utterances are never changed and always preserved.

    Generate AI-powered intents and utterances for multiple languages

    After you create intents and utterances for the first five languages, you can change the flow to the language for which to add utterances and perform the same steps with Virtual Agent to create them for each additional language.

    Note: When you generate intents with Virtual Agent, only utterances for the default language are created. If you change to another language, the language includes the intent, but you must generate utterances. The Intent Description cannot be modified from the default language description.

    Map slots at the intent level with Virtual Agent

    You can assign slots at the intent level. Slots appear in an expandable section on the individual intent’s page. You can add or delete slots. You can create slots that are not specific to a single utterance and instead applies across the entire intent, including when you change languages. When you delete the slot under Slot Assignment, Architect removes the slot for all languages and from any assigned mappings to utterances.

    1. Open the intent for which you want to add a slot.
    2. Click Slot Assignment. The collapsible view expands.
    3. Click Add Slot.
    4. Perform one of these steps:
      • To add a new slot, click Add Slot.
      • To assign an existing slot to the overall intent, click Select Slot and choose an existing slot.
    5. To assign a slot to a specific utterance, double-click the word in the utterance that you want to map to a slot and perform one of these steps:
      • Select the slot that you want to use.
      • Click Add new Slot and in the Add Slot dialog box, add the new slot, and then click Save.
    6. To delete a slot that applies to the entire intent, click the X next to the intent that you want to remove.

    Manage intents created by Virtual Agent

    In the Intents view, you can manage the intents that you create or that Virtual Agent creates. Go directly to a reusable task that Architect creates for a specific intent and edit the intent and utterances.

    Opt out of Virtual Agent

    If you revert your flow, Architect removes Virtual Agent functionality from your flow and you lose access to previously configured Virtual Agent settings. Then click Revert. You do not lose the intents or utterances that you already built.

    To downgrade your flow from Virtual Agent and revert your changes, export a previous version of the flow that is not Virtual Agent-enabled, and import it in your Virtual Agent-enabled flow.