The rise and rise of conversational AI: What is it and how does it work?

Automating routine and transactional interactions such as account management, order confirmations, returns, and address change with smart bots is on the rise.

The 1927 German expressionist movie Metropolis featured one of the earliest depictions of a machine communicating with humans. It has remained one of science fiction’s favourite subjects ever since. And artificial intelligence (AI) has now made it the billion-dollar solution to improved efficiency and reduced costs.

By 2026, conversational AI deployments within contact centres will reduce agent labour costs by $80 billion, estimates a Gartner report. It has also projected that one in 10 agent interactions will be automated by 2026, compared with the current 1.6 per cent.

Indian companies are increasingly adopting conversational AI across contact centres to assist human agents with real-time guidance. Automating routine and transactional interactions such as account management, order confirmations, returns, and address change with smart bots is also on the rise.

Uniphore Software Systems provides solutions to help contact centres cut the time and effort expended by human agents. The solutions aim at automating tasks like agent after-call work summarisation and call dispositions, using robotic process automation (RPA) for knowledge base integration, reducing the average handle time, and allowing agents to focus on more important tasks.

“One of our digital business customers in India deployed a solution that has reduced the number of chats handled by their human agents by 40 per cent,” says Annie Weckesser, Uniphore’s chief marketing officer. “This customer saw that their business was booming, and rather than hire new agents, scaled up with the rising number of chat-based transactions. In this case, our conversational AI assistant manages 70,000-80,000 chats per month, which were otherwise managed by human agents.”

Weckesser reckons companies can get the best results by deploying automation technology to help human agents, instead of replacing them. Reason: The automation process cannot eliminate human agents, as customers prefer human interactions over bots.

She adds that companies over the years have invested in applying conversational AI to customer self-service solutions such as intelligent virtual assistants with voice bots and chatbots. “However, people still want to speak with humans for their most important and personal matters. There is a huge opportunity to implement conversational AI in agent-assisted conversations, such as using it to drive real-time agent guidance based on customer intent, sentiment, and tone.”  

According to Gartner, labour expenses make up around 95 per cent of the total costs of contact centres. There are currently some 17 million contact centre agents worldwide, while many organisations struggle with rising labour costs and staff shortages, the study said.

Ashwani Arora, executive director at market research firm Market Xcel, says the adoption of conversational AI increased dramatically during the pandemic owing to higher demand and worker shortages. “Indian companies across industries started using smart bots rapidly to enhance customer engagement. The current market for conversational AI in India is $0.5 billion. It will grow substantially in the near future, with a growth rate of around 20-25 per cent.”

Elaborating, Arora says conversational AI will grow through newer applications.

“For instance, navigation of a banking website like HDFC Bank to obtain relevant information can sometimes be a time-consuming process and yield irrelevant search results. HDFC Bank introduced the electronic virtual assistant, an AI-powered banking chatbot. It is available 24/7 to help customers with their accounts, cards, or any other service,” he says.

The worldwide end-user spending on conversational AI solutions within contact centres is projected to reach $1.99 billion this year. It includes the budgets spent on solutions such as agent assist, automated knowledge bots, and training contact centre agents.

Rezo AI, an automation company in Noida, helps customers like Maruti Suzuki, Tata AIG, and ACT Fibernet resolve customer queries while maintaining a low average handle time.

Rashi Gupta, the firm’s chief data scientist and co-founder, says one of its customers, which operated as a third-party contact centre for Paytm, found that only one out of every 200 interactions by its human agents was productive for their business. “On average, a human agent working for eight hours could find only one person interested in their offer. On the other hand, we are doing 15 lakh calls for various customers every day using a bot. In other words, 200 calls, which used to take a day, can be done in one minute.”

Nevertheless, Rezo is working on improving interactions by adding a human quotient to the AI-driven conversation, Gupta reveals.

“The next disruptor in natural language programming will be technology that helps interpret customer conversations mingled with sentiment, emotion, accurate answers, and context, ending dead-end conversations. LaMDA, by Google, adds pieces to one of the most tantalising sections of that puzzle: conversation,” she adds.