According to Cathal McGloin, conversational AI is the latest evolutionary step in today’s technology.
As the CEO of ServisBOT, an AI-based customer service startup, he argues that conversational AI is central to of the democratisation of AI.
Computer interfaces, he says, have evolved significantly every ten years: in the 1960s, it was the punch card; in the 1970s, it was the green screen; in the 1980s, graphical user interfaces; in the 1990s, it was the laptop and in the 2000s, it was touchscreen.
“Conversational AI is the next evolution, coming after mobile,” he says. “And what’s driven the evolution to conversational AI is firstly that smartphones’ processing power is getting more powerful all the time, and secondly AI and its ability as a service is becoming more prevalent.”
Conversational AI lets the conversation dictate the process
The availability of AI is down to the technology giants, says Cathal, with Amazon, Facebook, Google, Apple open-sourcing their AI technology and making it available to the ordinary developer.
He describes how natural language processing (NLP) engines are the “language of AI”, providing a high degree of accuracy at understanding and interpreting what is said and the user’s intent. This differs from AI in its simple form, which decides what to output from what the input is.
“That can be predictive,” he says, “But it’s not high in accuracy.”
“We’re moving towards a world where people prefer to interact using natural language to get information out of things.”
“We’re moving towards a world where people prefer to interact using natural language to get information out of things. Say booking a cinema ticket or getting a pizza. You see it with Amazon Alexa and Google home devices, where we’re using spoken language, like say ‘Order me an Uber’.
“So if I come in and I say I’d like to book a hotel, it shouldn’t be a seven-step process. I should be able to tell it what I want. So conversational automation is that I give the data in any order I want and that dictates the process.”
The idea, he says, is that rather than automating a claims process or application process according to how the business wants to do it, conversational AI lets the conversation dictate the process and not the other way around.
Not just a “little chatbot, little logo in the bottom right-hand corner”
Cathal believes that as chatbots develop, they will ultimate drive change in how companies interact with their customers.
“We believe it’s going to have an impact in how enterprises need to engage with the customers and make things available to the customers,” he says.
“So we don’t see it purely as a little chatbot, little logo in the bottom right hand corner of the screen. We see this as a much bigger confirmational shift in how people will engage with the technology going forward.”
Cathal explains that conversational AI will go beyond the chatbot, it will be everywhere you go – on your website, your mobile or your social networks. It will respond to natural language, text, chat and voice.
“We see this as a much bigger confirmational shift in how people will engage with the technology going forward.”
“So, rather than you having to understand how to transfer money, you won’t need that you’ll just go on to a website and say ‘transfer money to my mum’.”
All sorts of technologies are needed to make this happen, and he says that is what ServisBOT is “orchestrating the chat equivalent of microservices”.
“So if I can build lots of microservices to check my balance, or reset my PIN, do this, do that, and then orchestrate all these around conversations, then we create the verbal equivalent of technical microservices.”
A microservice is a service-oriented software development technique, where services are fine-grained and protocols are lightweight, so that the application is easier to develop and understand.
“The key will be turning technical interfaces or APIs into conversational elements,” says Cathal.
Practical benefits of conversational AI
Cathal talks about insurance companies as an example of an industry that would use conversational AI. When a customer signs up, the business needs to see their driving licence or no-claims bonus and the ServisBOT solution would then send a bot out to the new customer.
He explains: “The bot says ‘Welcome to your new insurance, customer’ and says ‘Upload a photo of your documents right now’ and ‘Do you have any questions?’. And the whole experience will be a conversation driven one.”
One of the big problems he says ServisBOT has solved is reach: “In a world where people don’t take phone calls, from numbers they don’t know especially. How do you reach customers, in say a cash-claim business? Or if you have outstanding debts?”
For example, if a customer leaves their UK utility provider, the provider has to go through a process to allow them to close their account.
“Customers don’t want to take a phone call, but they will engage with a bot.”
“What happens in that utility company today is that at the end of the week, they give a list of 5,000 leaving customers, and they give it to a ‘smiling and dialling’ and say please try to save these customers,” he says.
ServisBOT created a bot which is sent out to those customers. It advises: “I’m here to help you leave”, which the company has to do, but then asks: “By the way, is there anything I can help with?”
The bot reaches the customer, says Cathal, “in a way customers want”.
“They don’t want to take a phone call, but they will engage with a bot or download an app. And it turns it into a potential opportunity to reengage with that customer.”
And the lack of human interaction here is used to make a positive experience.
“They don’t necessarily need it to be a human experience”
ServisBot is working with a utility provider in the US, which it cannot name. However Cathal stresses that the data coming out of the partnership demonstrates the bot’s benefits.
“We have proven that we can reach. Just in sending a message, SMS has 98% open-rate. We’re getting 80% plus opening, versus 3-5% answering calls.”
Whether the company can keep the customer then depends on the company itself and what it offers.
Banks could use the tech too, such as when people have questions around charges.
“There are opportunities to improve service at all touchpoints whether its claims, cash collection, on-boarding, retention and brands. Trying to reach customers across different channels. They’re trying to be there on Alexa, and they’re trying to control the experience customers have," he says.
“They don’t necessarily need it to be a human experience. But there will always be human involvement where it gets too personal or complex.”
The challenges for
When it comes to conversational AI in the wider technology world, there are challenges. But Cathal believes the technology will ultimately become pervasive.
“It’s a very noisy space,” he says. “People are using AI the same way that people used the internet 20 years ago, ‘Oh it’s the internet’.
“But it’s about how you use the technology. AI is a big word, it covers so many fields. It could be visual or speech recognition, deep learning and so on. The truth is AI will become a standard feature of all software technology going forward.
"If it doesn’t have machine learning or some aspect of AI in it, it will fail. In the same way no one asks if there’s a database, no one will ask if AI is involved, it will just be obvious it’s there.”
Images courtesy of Brolly