Chatbots, a new way of serving customers

Informatics R&D

In recent years, chatbots have been perfected and made their way into day-to-day tasks related to customer service. They ask questions, they answer, and they converse to make our life easier, but are they really as effective as they seem? They offer many advantages, although there are still some problems to solve.

Chatbots are programmed to interpret the reason behind the conversation and to answer questions.
Chatbots are programmed to interpret the reason behind the conversation and to answer questions.


A chatbot is a software application based on Artificial Intelligence that can have a conversation in real time by text or voice. In the first instance, there are the text-based customer service chatbots found on banking, insurance, travel, catering, etc. websites. Then we have the famous voice-based virtual assistants —Siri, Irene, Cortana and Alexa— designed to try and make our lives easier by answering our questions.

And they're not magic, they're science. Chatbots are programmed to interpret the context of the conversation and answer questions using Artificial Intelligence tools such as Natural Language Processing (NLP), which enables them to understand us, or Machine Learning, with which they can automatically learn from each interaction.


Depending on the use they make of that Artificial Intelligence, chatbots can be one of three types:

 Linear conversational
These are based on a decision-tree architecture and are not very smart. Their flow of answers is determined by a linear chain of stages, so they give automatic responses rather than establishing a fluid conversation.

 Non-linear conversational
Thanks to Machine Learning and NLP, they can interpret the user's intentions and the context of the conversation in order to respond accurately. If a user makes a request, the bot will understand it and return a few options tailored to his or her requirement.

 Hybrid conversational
A combination of both the above that enables fluid and personalised conversation with users. If the chatbot doesn't know how to answer a question, it immediately notifies a human agent to add the answer into its database.


Although they seem like a recent invention, chatbots go back to the 1960s.Eliza was the first. This was a rudimentary software program created in the Artificial Intelligence (AI) laboratory at MIT. It could simulate a conversation by using a pattern comparison methodology. One thing that is certain is that the chat was more monologue than conversation.

Despite its limited functionality, Eliza fired the starting gun for a dizzying race that has led to today's chatbots, which are far more intelligent and sophisticated. This qualitative leap has been so great that many companies are adopting chatbots as a new, fast, efficient and profitable way to provide customer care services.

According to data published by the Gartner consultancy, 38% of companies worldwide plan to use conversational bots as part of their customer care service. Furthermore, their use will be widespread in less than five years, not only in this area, but also in terms of generating sales and fostering customer loyalty. So, what exactly is a chatbot?

Steps required to implement a chatbot.
Steps required to implement a chatbot.


The use of chatbots, as pointed out by the Gartner consultancy, will soon be generalised in the field of customer service, although not only in this field. Nowadays, virtual assistants are beginning to be used in companies to promote sales, retain customers and generate content of interest.

Chatbots have numerous advantages: 24x7x365 availability, fast and efficient response, constant learning, cost savings, easy personalisation and even social media management! And the disadvantages? Although natural language processing is getting better, these tools don't capture certain characteristics of human language such as double meanings, sarcasm and moods, which can lead to customers getting frustrated. Also, many people, especially the elderly, reject these types of technological advances and there is the possibility that the system may fail mid-conversation, causing discomfort and mistrust in the user.

 Virtual reality, another world at your fingertips

 Data Mining will predict your behaviour