When a user sends a message in a non-English language, our Translation layer helps our ML model understand the user's intent and go to the desired step to fetch the response in the user's language. Our Correction layer on top of our translation layer helps us to achieve the desired results.

So basically, when a user enters a query in a non-English language, the stages that the bot would go through are:

• Translate that sentence in English.
• Understand it, and accordingly, look for the step closest to the user's intent. 
• Send a response to the user from that step's bot response. 

User Messages

Let's understand this with an example.

• In the image below, the user asks the bot for help in Spanish, by typing Hola necesito ayuda.
• The bot would auto-detect the language, and would translate this sentence into English.

• After translating the Spanish text in English, the bot directed Help-related responses to the user's chat, since the translation in English triggered the Help step.

Bot Response

Now, once the Bot has understood the user's intent post-translation, it's time to send a response to the user. 

Bot Response plays a very crucial role in the Multilingual Bots. 

• In Bot Response, you need to add responses in all languages that your bot supports.
• If your bot supports Spanish, French, Arabic, etc then, you would have to provide responses in all the above languages.

• In the above image, as you can see the bot response has a response both in English and Spanish, and also the languages that are supported.