A multilingual smart chatbot (also known as Conversational AI or Intelligent Virtual Assistant) is an artificial intelligence based machine that understands what users ask. Essentially it is a machine that presents information to the user in the language of the user.
So if you are using a chatbot in a multilingual mode it means that the answers of the chatbot are in the language of the user. Meaning that if the user speaks English then the answers and buttons of the chatbot are in English. If the user speaks Spanish then the answers of the chatbot are in Spanish as are the buttons.
The language of the chatbot can be determined either manually or automatically. The manual choice is simply giving the user the option to choose the language before interacting with a chatbot.
The automatic option is to present the chatbot in the language of the website that the user is browsing. For example, if the user is on a website and has chosen the language of the website in Spanish then the chatbot will give answers in Spanish. If the user is on the website and the selected language is German then the chat bot gives answers in German.
In the case of these smart AI chatbots that are capable of understanding natural language, the language detection can be built in. What this means is that if a user comes to the bot and starts chatting in English then the bot will answer in English. However if the user comes and starts asking questions in German then the bot will answer questions in German. You can try a multilingual chatbot here.
Such multilingual chatbots are used in customer service. The reason is that a chatbot in different languages suits well for companies that have to provide support in multiple languages.
In cases where you have to support several languages and lack native speakers of the language, a chatbot that is capable of providing support in various languages automatically, is highly appreciated.
The first main benefit is that you need less multilingual support staff. People who speak several languages fluently are usually difficult to find. As such utilizing a multilingual chatbot means that you no longer have to search for extra multilingual staff.
The second benefit is that your users like to be served in their native language. Most people prefer to speak to customer support in their own language because they can better express themselves in that language.
So when you offer support in their native language, your customers will most definitely appreciate it and be happier with your company and the products and services you provide.
For creating multilingual chatbots you have essentially three options. We will break these down in more detail in the chapter below but here is a summary outlining what each of the options are and their pros and cons.
With AlphaChat you can use this when your bot has few intents (i.e. topics that the bot knows). There is no exact number as to what is few but in general if you have less than 15-20 actions then you should be fine in having multiple languages in one bot.
What you need to do is create a topic for the bot in one language (e.g. English).
1. Add a topic.
2. Give the topic a name.
3. Add an answer to the topic.
4. Add phrases and utterances to the topic. 10-15 phrases will do.
5. Train the topic.
6. Test the topic with some of the phrases that you just used for training.
For more information we have added a video below where you can get an overview of how to add text, images and links to the bot content.
And here is a video explaining how to add training phrases to the chatbot.
Now when you want to add the same topic in another language (e.g. German), then make a new topic and go through the same steps as above but only add the topic in German. We suggest to create a folder for each language and to put all topics in that language under that folder. This makes handling your bot content much easier as you add more languages.
1. Create a new folder for German language and add a topic in German.
2. Enter the answer in German for the topic.
3. Train the German topic with German phrases.
4. Test the bot in German.
Now you have a bot that can answer questions both in English and in German.
It makes sense to make a separate bot for each language once you have more than 20 topics. The reason being that if you see that over time (in the next few months to a year) your bot will have 50+ topics then making a separate bot for each language makes sense.
The sheer number of topics will make managing the content a bit cumbersome. In addition there might be overlap between topics between several languages if you have many topics.
For example if you offer a credit card titled Mastercard Gold for your customers, then that name can be the same in English, Spanish and German. So when you only have one bot and customers ask a question in Spanish then the bot can answer in German. This of course is not good for you and your customers.
When you have few topics and little overlap in terminology about your company and products, then having all the different languages in one bot can work. However, when your bot grows then it can start providing wrong answers.
If you wish to create a separate bot for each language then the steps to follow are the same as in the example above. The only exception being that you create the topics in one language only (e.g. only in English and not in any other language).
For single language bots we suggest creating a folder for major topics (e.g. credit cards, withdrawals, loans etc if we continue our banking example). You will have a variety of topics under those larger umbrella topics and it is easy to keep your content organized.
If you have more than 15-20 topics per bot but you still do not want to go through the hassle of creating a bot for each language separately then you can use machine translation with AlphaChat.
How that works is that you create a bot for one language, let’s say English. Now when a customer comes and asks a question in Spanish then the Spanish phrase is translated into English. The English phrase is fed into the bot AI engine where the AI determines the topic and provides an answer in English. The English answer is translated into Spanish and the Spanish answer is shown to the customer.
All this is done in less than a second and all the customer sees is a question she asked in Spanish and the answer the bot gave in Spanish. The number of languages available depends on the translation API that is connected with the bot but using either Google or Amazon there are 50+ languages available.
Here is what you need to do if you want to have one bot that knows all languages through auto translate.
What you need to do is create a topic for the bot in one language (e.g. English).
1. Add a topic.
2. Give the topic a name.
3. Add an answer to the topic.
4. Add phrases and utterances to the topic. 10-15 phrases will do.
5. Train the topic.
6. Enable the machine translate feature and select which translation API provider you wish to use (Google Translate or Amazon Polly).
7. Test with English and with German phrases. The bot answers both though it is only trained in English.
Multilingual chatbots are interfaces through which users can chat and get answers to their questions in several languages automatically.
You have options for building multiple languages into one bot if your bot has few topics. You can also create a bot for each language separately or you can use machine translation for making a bot in one language and have it available in all languages.
With AlphaChat you can create your own multilingual bots in all three options. Feel free to play around with the product and if you have any questions, feel free to reach out.
