Multilingual Support - Introduction

Enabling Multilingual Support

To enable multilingual support, you must add the desired languages to the Supported Languages section. Follow the steps below to configure this feature:

1. Navigate to the Configure Section: Access the Floatbot dashboard and go to the Configure menu.
2. Select the Multilingual Option: Within the configuration settings, choose the Multilingual tab.
3. Add a Primary Language: Set the Primary Language, which will serve as the default language for the AI agent's initial interactions.
4. Add Supported Languages: Use the Supported Languages dropdown to include additional languages that the AI agent should understand and respond in.
5. Click the Update Button: Save your changes by clicking the Update button to enable multilingual functionality.

Working with Workflow for Multilingual AI Agent’s

Once multilingual support is enabled, the next step is to implement language selection within the workflow. This ensures that the AI agent can effectively interact with users in their preferred language. The process begins by prompting the user to choose a language, and once selected, configuring the system to handle interactions in the requested language.

Example Implementation:

Suppose the AI agent supports both English and Spanish. Here's how you can structure the workflow:

1. Prompt Language Selection: Begin the conversation by asking the user to select their preferred language.

Example prompt: "Please select your preferred language: English or Spanish."

2. Set the ASR (Automatic Speech Recognition) Language: Once the user selects a language, configure the ASR and NLP (Natural Language Processing) components to operate in the chosen language.

If the user selects Spanish, the ASR should switch to recognize and process Spanish inputs.

3. Redirect to the Appropriate Workflow: Based on the selected language, route the user to the corresponding language-specific workflow.

• In this example, selecting Spanish will transition the user to the Spanish workflow, where all subsequent interactions will be handled in Spanish.

This approach ensures a smooth and personalized user experience, aligning the conversation flow with the user's linguistic preference.

As demonstrated in the video below, the workflow begins by prompting the user to select a language using the User Input item. The selected language is then identified through Intent and Entity Detection, please refer section below to understand how to detect the language input.

Following this, the workflow branches based on the user's language choice, as shown in the conditional logic setup:

• If the user selects English, they are redirected to the English workflow. Additionally, the Change ASR item is used to set the ASR system to English.
• If the user selects Spanish, they are directed to the Spanish workflow, where the Change ASR item configures the ASR system to operate in Spanish.

This structured approach ensures seamless language handling, enabling the AI agent to deliver an optimized experience tailored to the user’s selected language.

Working with Intents for Multilingual AI Agent’s

Now, let’s explore how to leverage Intents for multilingual support, starting with detecting the language selected by the user. Follow these steps:

Step 1: Create an Intent for Language Selection

Begin by creating an intent that captures the user's language preference. This intent should include a variety of language selection phrases to ensure accurate detection.

• Example Variations:
  • "I would choose English."
  • "Spanish is preferred."
  • "Please set my language to English."
  • "I’d like to continue in Spanish."

These variations help the AI recognize different ways users might express their language preference.

Step 2: Create an Entity for Languages

Next, create an Entity to represent the available languages (e.g., English, Spanish). This entity will act as a reference point for extracting language information from the user's input.

Step 3: Associate the Entity with Intent Variations

Link the Language entity to each variation within the language selection intent. By marking the relevant part of each input (e.g., "English" or "Spanish") with the entity, the AI system can accurately extract the user's chosen language.

This setup ensures the AI agent can reliably detect and process the user’s language preference, laying the groundwork for seamless multilingual interactions.

As seen in the video below, the user created an intent, defined an entity, and then marked all variations by double-clicking on the keywords.

Finally, we will learn how to configure FAQs or set up redirections based on language selection for a multilingual AI agent. This ensures that users are presented with the appropriate responses or workflows tailored to their chosen language.

As shown in the video below, two separate lists are displayed for each language. If you want to respond with text, provide the relevant text response for both the English and Spanish intents, as demonstrated in the "User is asking who are you" example.

For flow redirection, ensure that users are directed to the appropriate language-specific workflow, as illustrated with the "Harassment" intent. This approach guarantees that responses and actions are aligned with the user’s language preference. The decision to detect the correct intent is handled by the NLP engines. When you select Spanish as the language and ask a question in Spanish, the query is first translated into the primary language. Then, this translated query is processed to detect the correct intent. Finally, the response is returned in the original language of the user's input.

Note: Do not forget to save the intent separately for each language. If you make changes to the English intent, save those changes, and then save the changes for the Spanish intent separately.