Chatbot and Artificial Intelligence

Artificial Intelligence (AI) or Natural Language Processing (NLP) takes your chatbot to the next level! 

You can teach your chatbot new and new words and sentences so he can communicate with your customers on the highest level possible and send them the information they are after.

In the Brand Embassy Chatbot Builder, you can create such flow with the AI Intents.
Just remember, the AI can learn only one language ;)

The way it works is that you give the chatbot variety of sentences which he checks when the customer messages and sends him into the flow accordingly.

Let's have an example:

Let’s say you are running an ice cream stand chain. You have a FB page where people message you a lot. You have noticed that the customers ask mainly about two things. How much do various flavours cost and where to find the nearest stand. How to make the chatbot recognize these questions?

First, let’s add few phrases that customers use when asking about either of the topics.

Now let’s try if the chatbot understands us if we ask where we can buy ice cream?

Almost 94%. That’s pretty confident, chatbot would safely understand this.

Now a customer complaining about the price of vanilla ice cream arrives.

In this case, the chatbot would not recognize the Intent at all. What now?
Let’s add a few more phrases to the Intent settings and see if the chatbot can understand this better.

When the very same phrase is tested, you can see the additional sample phrases increased the chatbot’s confidence.

Once you have the Intents ready, set up the auto-response to send the customer to the correct flow.

In the auto-response settings select trigger type between intent recognition or trigger words. Then define specific intent or words, which you desire to act upon.

If the confidence rate is between 60% and 80%, the customer will have to confirm the validation question before the response triggers. The customer can answer either Yes to confirm their intent so the chatbot continues with the auto-response, or No and the chatbot continues with the Fallback response.