Creating a Humanized Conversation in Banking

Since Facebook announced that its Facebook Messenger platform is open to chatbots, we witnessed a huge proliferation of different kinds of chatbots, operating on different platforms, to perform different functions in different sectors. However, it is apparent that the majority is rule-based, question-answer type that lack the dialog flow structure and the ability to continue a conversation in a humanized manner.

All the AI-based chatbots do not necessarily provide a humanized communication, although they understand the natural human sentences. A humanized conversation is far beyond classical question-answer or request-action kind and it requires different technical features. So what are these features that make a chatbot humanlike? In this blog, we will point out some of these features through a banking chatbot. The reason we choose a banking chatbot is that providing a humanlike experience is very crucial in this sector and banks prioritize this characteristic.

Why is humanized conversation important for the banks? 

As banking is a centuries old industry, the journey of customer experience is a long way from the old days of customer representative at a branch to the AI-powered chatbots or virtual assistants mimicking the former via instant messaging. 

The pre-internet era: In this phase, the customer representative, providing the Bank a human face as the prime actor of customer interaction, gives the customer only the information that is requested and relevant to that specific situation. 

The internet era: Bringing web sites in the scene means interacting with our banks by clicking on links, menus, and buttons, and filling out forms. This is definitely complex, hard to use and provide limited personalization.  

The mobile era: In the subsequent period, banks developed their mobile apps. However, this created a world where the user has to download and manage an app for every tiny function and it became difficult to manage it for the users as well as the companies. 

The conversational era: Now, we are in the age of “conversational” which provides the closest experience to the one we used to get in the branch while talking to our customer representative. “Conversational banking” is the most natural, direct and easy way of banking. Unlike the self-serving web and apps, the chatbots and virtual assistants provide a service similar to the banking customer representative although they are also self-service channels. As the customers need to communicate with the “human face” of the Bank and have the perception that “somebody helping them”, conversational banking has the potential of re-humanizing the banking experience. Nevertheless, the crucial point is that the chatbot has to be designed in such a way to ensure humanization in both technical and CX aspects. 

The features that make a banking chatbot experience really humanized

These are the set of features that enable a banking chatbot understand the customers with high accuracy and remember what they have asked 2 minutes or weeks ago. This kind of a banking chatbot can extract all the entities of an intent from a single sentence or collect the missing input and make sure that the conversation flows as if the customer are talking to a human. Before defining each feature, we need to define the terms “intent” and “entity”, an intent is the user’s intention. For example, in a sentence like “transfer 100 $ to dad”, the intent is “transfer money”. An entity is the information that modifies the intent. In the same sentence, the entity is 100 $. Amounts, dates, months, years, interest rates, currencies are defined as entities in a banking chatbot.

High accuracy NLP engine: The primary prerequisite is that the NLP technology behind the chatbot should be advanced to provide a high accuracy in the banking domain as well as in the specific language that the chatbot serves. This is crucial to make chatbot first understand what the customer tells and provide the most relevant response. 

Intent follow-up: This feature enables the banking chatbot to follow a conversation just like a human would do. It tracks the existing intent and allows user to make updates or ask specific follow-up questions in a topic. This feature allows such a conversation to take place: 

User: Send 100 USD to dad tomorrow.

Chatbot: Sending 100 USD to dad tomorrow, do you confirm?

User: Make it 200

Chatbot: OK, sending 200 USD to dad tomorrow, do you confirm?

Bots, that do not have this feature, cannot process “Make it 200” input and requests user to provide all the input again which greatly diminishes the experience.

See the video here.

Automatic intent change: This feature allows the user to change topic of the conversation and get back to the previous topic if needed. It detects that the input is off the topic and automatically changes it. This feature allows such a conversation to take place: 

User: Send 100 USD to dad tomorrow. 

Chatbot: Sending 100 USD to dad tomorrow, do you confirm?

User: How much do I have?

Chatbot: You have 400 USD in your account.

Chatbot: Would you like to continue sending money?

See the video here.

Chatbots, that do not have this feature, require special keywords to change the topic which would not occur in a human-to-human conversation.

Conclusion

Cbot, a leading conversational AI company, invests a lot in making its banking chatbots and virtual assistants more humanized. These 3 features that the company offers to its customers, support Cbot’s product to be more powerful and preferred by the banks. Cbot continues to invest in R&D to put its banking virtual assistant in a higher position both in terms of the technology and the customer experience it provides.