A number of brands joined the bandwagon with their own bots for Messenger, allowing users to interact with businesses and solve basic queries without requiring human interference. Even as the chatbots evolved to solve more complicated queries like financial transactions, its limitations triggered a new demand for AIs that are smarter and can understand complicated non-linear human queries.
Since Chatbots’ debut, newer AIs have come into the picture. And businesses around the world are increasingly integrating them into their systems. The primary reason behind the growing need for smarter and improved conversational AIs is the rapid digitization of the world where businesses are required to assist customers 24x7.
While the rudimentary bots can do a basic job, to address the complex human queries, AIs need to be smarter. Over the past few months, these conversational AIs are not just automating customer support but also gaining more insights on how to personalise and improve services for individuals.
What is conversational AI?
Conversational AIs are essentially an evolved version of chatbots and aim to conduct human-like interactions with humans. The AI uses natural language processing (NLP) to understand how a normal person will communicate, instead of trying to serve a pre-defined set of options.
For instance, a customer can ping a business anytime of the day and raise a query. With an intelligent system in place, a machine provides an answer to the query, in a manner that a human will do. A text-based conversation allows customers to keep their interactions at one place without disrupting their day-to-day work. The key difference is that the conversational AI is capable of understanding the complex structure of sentences which may not necessarily be linear.
New age communication
The spectrum of a conversational AI is becoming broader. From basic question and answer, the AI can provide customers insights or walk you through processes. And in return, businesses can leverage these data sets to identify what their consumers are looking or expecting from them.
Modern conversational AIs are looking beyond the text-interfaces to voice, making the experience richer and easier.
A great example of a highly capable conversational AI is Google’s Duplex. The AI mimics human voice to make appointments on users’ behalf. The Duplex AI, however, isn’t being rolled out to enterprise consumers. Instead, Google is reportedly working on a new AI system called “virtual agents" for businesses.
“Designed to work with existing contact center technology, Contact Center AI makes it easy to train an AI model to interact with customers and provide insightful direction for agents. The result? A more personalised, intuitive customer care experience from the first ‘Hello’," Google explains on its website.
From IBM, Cognizant to SAP, a number of IT companies are building and offering conversational AI solutions.
The reception, adoption
Contrary to popular belief, users are increasingly getting accustomed to these conversational AIs. While traditional channels for interacting with businesses continue to exist, consumers are also aligning themselves with the new-age tools. The adoption, even though at a nascent stage, is increasing in India.
“Enterprises are very quickly adapting the technology… they are looking for solutions like this… because there are big challenges they are facing," says Sirdhar Marri, founder and CEO of Senseforth.ai, an AI-based bot platform.
“As enterprises, their growth and scalability are at an inception point because they have to invest more on human resources to scale up the kind of growth they’re expecting. Now, that is a big challenge. There are not enough trained people and even if they are there, businesses have to factor in things like training costs. So, what they want to do is optimize on the human resources. Say, a company has 3,000 people for customer support, and now these 3,000 can support x amount of customers and scale up to 2x of the total customers. The business driver, however, will come from learnings like how do I scale my interaction with a customer and how do I make it engaging? How to make the support available across all kinds of platform the user is coming from?" Marri said.
“Typically what happens in our case is that our solution is pre-trained with human intent and they go live after four to six weeks. If a solution is complex, the AI may take 8–10 weeks. Now, once it goes live, we start with select users. Once they start using it, bots give us information as to what the users are asking and where the AI is able to answer properly. We also monitor where AI performance is high and where it is low. All these data is sent by the bot to us. Now, we will supervise the learning of the bot in the first three months. The first month is where focused group rollout happens. Supervised training of the bot happens for almost three months," he said.