Now we come to the practical applications that may interest small and medium-sized companies that work on the internet in everyday life.
One of the most innovative uses of these practices is certainly the recognition of language. Thanks to powerful computers and the massive use of Deep Learning, some platforms provide web-based applications that offer services that a few years ago were exclusively at the service of a few large companies: conversing automatically with a client.
We have not yet reached the conversational ability of Samantha, the female voice of the computer that Joaquin Phoenix falls in love with in the highly successful film ” Lei “, but we can surely exploit this technology (commonly called Natural Language Processing ) to our advantage.
We can therefore automatically solve the needs of the customer who turns to our customer care without the intervention of an assistant in the flesh, but using sophisticated patterns (models) of response based on the customer’s request.
Chatbots: we talk to the customer
We have the technology, we have the tools to understand what the customer is telling us but an essential tool is still missing: the interface with the customer .
The most used tool to communicate with the customer on the web has always been e-mail. We all know how expensive the client-company relationship via e-mail is: expensive in terms of time, money and above all not as effective as a chat, for example like the ones we usually do on WhatsApp.
Fortunately we have the solution, the chatbots. What are chatbots ? For a more in-depth explanation I refer you to one of our articles on the subject , but for the moment know that they are virtual users of a chat that respond to the user thanks to their Artificial Intelligence. The chatbots unfortunately cannot be used on WhatsApp (they are banned as soon as they are identified as such by the platform) but this does not mean that they cannot interact with customers on our website thanks to specific graphical interfaces called plugins.
Certainly it will have happened many times, within a site, to click on an icon with a text like “chat with us”. This is a graphic plugin, which allows you to open a small chat window with a sales representative. I tell you a secret: many times, especially at the beginning of your conversation, you are talking to a chatbot, who understands the tone and the subject of your request, addressing you to the most correct Customer Care agent available at that time.
For those who want to try to build their own personal agent, let’s try to explain briefly what are the fundamental steps:
- Needs analysis: in this phase all the possible requests to which the agent must respond are identified and the available resources are identified (eg FAQs already prepared)
- Implementation on a NLU platform
- Test on a panel of users (internal and / or external)
- Refinement of the User Experience through user feedback
Obviously all this is very complex and is usually done by developers specialized in the field, but with a little time and, above all, a lot of perseverance you can manage to create a chatbot starting from the documentation and tools provided by the most advanced NLU platform (according to me): DialogFlow.
In fact, the Google creature, thanks to a convenient web interface, is able to provide us with all the tools to create a correct flow of interaction with the customer.
Furthermore, the convenient Training function allows you to refine the skills of the agent thanks to the supervision of a human.
Note the automatic integration with the Google Assistant: by clicking a simple button we will have all the power of textual recognition and the natural language of the new Mountain View creature.
The advantages of Virtual Assistants through chatbots
There are thousands of web pages that already explain in detail the advantages of this approach to customer care, so we will simply list a few:
- Scalability : is our customer care office no longer able to respond to all user requests? The answer is not to hire more staff but to automate the demand analysis process as much as possible and respond accordingly.
- Efficiency : stop the exhausting email exchanges; the customer no longer has to wait for nine in the morning to get an answer, the service is available 24/7
- Modularity : created an agent, it will be enough to implement a few more lines of code to satisfy the new needs, generated for example by a new product that you want to launch.
- Multilingual : up to 25 languages available.
- Marketing automation: once the user has chatted on our Facebook page we can send targeted promotions to users based on their requests.
- Integration with Messenger and the other chatapps: those who chat with our virtual assistant have become our Messenger contact (or Facebook if you prefer) and we can interact with him with all the tools that Facebook provides us (GPS position tracking, sharing of data and photos, voice messages, Messenger campaigns, etc). Let’s not forget that the same can be applied to Telegram, Skype and many other messaging platforms.
This was only a quick digression on the topic of Artificial Intelligence and its concrete applications. There would be much more to look into and study. If you have any doubts or would like more information on a topic that interested you, leave your comment, I am at your disposal.