Automation and AI: Trends for 2020 by Gianluca Pereyra
Our co-founder – Gianluca Pereyra – was one of the speakers at Insights by Visor.ai event and he went back a few years to explain who we are, how everything started and where we want to go whilst automating communications between companies and people.
Lisbon City Hall case
The year was 2016 and the Lisbon City Hall managed to open to the public a set of challenges the organization itself had and was trying to solve.
One of the challenges that drove our attention was that Lisbon City Hall wanted to improve the relationship they had with Lisbon citizens. Their contact center dealt with innumerous repeated interactions through email, phone calls and social media – out of curiosity, Lisbon City Hall Facebook page has nearly 500k followers.
We understood the problem and tried to analyze which questions were arriving at the contact center and what made people come and talk to Lisbon City Hall. What was the conversation about?
So the first step was to automate and drive the contacts to the right channels. For example, contacts about the promotion of events were forwarded to the Marketing department and people who wanted to complain were directed to customer support.
What started with just a conversation has become a business with more than 20 use cases delivered
They were very satisfied with the results and they asked us: “Can we do more? Is automation enough? What is the automation you’re providing us?” From Wikipedia ‘automation is the technology by which a process is performed with minimal human assistance’ and you don’t need necessarily to apply AI in automation. We automated the conversations and the question came: “Now, can you give an answer to the people living in Lisbon? Can you go a step forward?”.
The answer and simultaneous question that come to people’s minds is “AI” and “What is AI?”. ‘AI is the development of computer systems able to perform tasks that normally require Human Intelligence’ (The Oxford Dictionary). We are talking about text recognition, image recognition, and speech recognition and how to apply AI to the real world.
From the different AI branches that exist, we took on Machine Learning and Natural Language Processing for Portuguese in Portugal to help Lisbon City Hall to speak better with the citizens.
What started in 2016 with just a conversation with Lisbon City Hall has turned into more than 20 use cases delivered over the past 3 years.
Trends to 2020
What are the trends for 2020, what will we see in companies willing to embrace business technologies regarding any sub-branches of AI?
We will find more companies ready to receive AI solutions. That means that we will have a well-established person in the company that takes full ownership. Furthermore, people are more and more receptive and capable of talking and understanding innovative AI solutions. And even though the legacy systems for the new technology not always simplify the integration process, banking and insurer are a good example of how this reality is changing (for example, Banco Best and Tranquilidade.)
Now, where will it be applied?
There are four different points that every big company that sells services or goods has:
Acquisition means increasing the number of new clients. Retention stands for ensuring and providing for the needs of the client through customer support. Profitability is the increase of the revenue that a client generates by understanding different patterns along with the clients and see where they can upsell through machine learning. And optimization is wanted for all the companies that have a lot of volume in operations and need a well-oiled structure.
We also know that AI solutions require some maintenance. So, as a 2020 trend, we would like to include our efforts to try to decrease this gap by having more low data maintenance solutions.
Where are your clients? Where are the people you need to talk with?
The website is usually your business card. Clients arrive there and they need to understand the information they see. They might need to talk to someone and you have to give them the opportunity to do it. It can be a chatbot, but also an email address or a phone contact, but you need to be ready and available to the people who already show interest in what you’re doing.
Website chat is an option, WhatsApp as well. Visor.ai already works with banks that want to use WhatsApp to make transactions. They chose WhatsApp because they realized that their customers use this channel more than any other. You have to understand where your clients are and where to connect with them.
Emails have been here for a few decades. When it comes to sharing documents, email is one of the preferred ways. How can AI be used in emails?
Emails often come with lots of paragraphs, lots of text, so it’s not necessarily easy to give a direct answer. By applying AI, we try to understand the subject or pattern in emails. This way it can be directed to the correct department or, if it is detected that some information is missing, it is automatically communicated and the client can reply with the required data or document.
When it comes to voice assistants, as long that the input is well received, it’s possible to answer through voice to a question. For instance, there are very big companies with a very good job of voice recognition. We’re talking about Google Assistant, Alexa or Siri that have hundreds of engineers working on speech recognition.
So, what they are doing is opening their virtual assistants for apps to be integrated there. That is to say, because people are using more and more the voice assistant.
First, in order to interpret the text, the noise needs to be removed. As a result, there are a lot of companies working on that part too. Then, the voice bot transforms speech to text. The text is analyzed and answered out loud.
Can we talk about empathy?
Usually when you want to see innovation in technology, and in this case, AI branches, first you need to see Asia. In fact, they do everything there but sometimes European are not aware because probably it is not in our scope.
In Asia, they are testing the behaviors and feeling recognition by trying to teach children remotely. They put cameras in front of the room and analyze the children’s faces. The teacher, in the distance, receives real-time information about the students. Which is playing with the phone, which might be showing faces of confusion and who’s most engaged? To sum up, all the different pattern recognition, so the teacher can then intervene at the right moment. Once again, an empathy bot future is all about learning to recognize different patterns in a crowd.