The heralding of AI within the CX sector promoted predictions of an all-robotic workforce, where human customers and artificial intelligence would communicate seamlessly, and contact center agents would be required for only the most complex of tasks.
Up to now, that hasn’t happened.
AI, it appears, has lagged behind the industry’s hopes and it remains a long way off providing the full spectrum of nuanced understandings that a human being, fluent in a language or not, can readily provide.
Almost two years ago to the day, Nearshore Americas published an article that said using automation in CX was “still a work in progress”.
But the wheel is slowly turning and with every month, AI’s ability to ‘understand’ human intention is being sharpened.
The pandemic propelled this change further and, with heavy hitters entering into the space, heightened chatter around the still relatively young technology. In April of this year, Microsoft pushed ahead with its strategies for healthcare with the acquisition of conversational AI developer Nuance Communications, while in June, Cognigy, a provider of conversational automation solutions with a low-code approach, raised $US44 million to fund scaling intentions.
Interest in intent analysis and conversational AI is “growing in spades,” Melissa O’Brien, Research Leader at leading global services analyst company HFS Research told Nearshore Americas.
“The last year has just accelerated the uptake because when Covid hit a lot of processes and interactions were disrupted, and companies didn’t have the staff to manage calls.”
The pandemic “exposed a lot of broken processes” within the CX sector, she said.
But conversational AI is now approaching a level whereby customer intent is understood well enough to begin to play a serious part in the CX journey and provide cost-efficient customer service across industries.
Optimization is Key
Conversational AI is constructed with a myriad of technologies. Bots built for conversational AI should be able to understand human intention through communication on the phone or through text-based channels, and imitate human interactions to provide a solution to a customer’s problem.
While this sounds straightforward enough, the reality is complex.
“Intent analysis, or intent recognition, is really one of the fundamental building blocks if you want to create any kind of intelligent automation of conversation,” said O’Brien. “Depending on how far you want to develop the natural language processing (NLP) element, that ability to be put into an interactive voice response (IVR).”
With the potential to reduce overhead expenditure by delivering automated customer service at scale, the potential impact of conversational AI market’s is being recognized. According to a Deloitte study published before the pandemic, “the early benefits from the adoption of Conversational AI mean the global AI-derived business value is expected to grow by an average of 30% annually.”
7.ai is one Nearshore player that is leading the efforts to increase the presence and power and the technology within the CX industry. The company’s SaaS platform, the Engagement Cloud, claims to be “the most advanced conversational AI platform”, and is based upon the company’s long standing contact center experience. The idea is that by blending high-level technical expertise with decades of operational experience in the call center space, the company can provide a tool that meets the needs of modern contact centers.
“We built the Engagement Cloud by bringing many different components into the company,” said Murtaza Khuzaima, solutions architect at 7.ai. “Using a build-from-the-ground-up chat solution, an IVR solution and a virtual agent solution that we acquired, we then augmented our models with our learnings.”
Khuzaima says that the company’s efforts in deepening the insight of its bots, as well as parallel improvements in AI technologies, have allowed the company to build a CX solution that has real impact on call efficiency and customer satisfaction. However, AI isn’t displacing agents yet, and the human agent remains an essential part of the customer journey. But major steps are taking place.
“Though the AI is good, it’s not quite there yet and there’s still a need for the AI to HR hand off. But even when the hand off happens, it happens with context. The agent can see what has occurred up to that point and doesn’t need to begin all over again with the customer,” he explained.
“The adoption of Conversational AI mean the global AI-derived business value is expected to grow by an average of 30% annually” — Deloitte
Even from errors, the understanding of bots can be improved. “When the customer asks the bot something that it can’t handle and the conversation is escalated to an agent, the agent can then highlight the utterance to explain what the customer was trying to say. That learning goes back into improvement of the models,” Khuzaima said. “We use actual conversations to train up the bots, essentially crowdsourcing the agents in a concept called Collaborative Tagging.”
The company’s efforts are paying dividends for clients willing to invest. According to data provided to Nearshore Americas by 7.ai, one of its clients – “a very large bank” – have seen a 27% improvement in customer resolution within the IVR, and an estimated US$1 billion saved over 10 years, through the use of the Engagement Cloud.
The containment rate – meaning the percentage of calls that are resolved by the bot alone, starts at about 33% for most clients, said Khuzaima. With time comes learning and improvements, with additional benefits like identifying the root cause of return callers.
Conversational AI Adoption
Improvements in conversational AI and the role of the pandemic in reducing consumer’s barriers to digital channels have prompted questions about the future of the contact center and the human agent. But for Ian Jacobs, VP Research Director at Forrester, conversational AI may generate as many questions as it provides answers for, for the running of a contact center.
Jacobs points out that this can have an impact on agent scheduling and company hiring efforts.
“If I’m running my own contact center – badged employees, not outsourced – and somebody tells me they will build a chat bot that answers some easier questions and over time gets better, I’m hearing that my labor requirements and measurements of quality will be changed,” he said.
Conversational AI has so far gelled well with digital teams that do not have a massive volume of interactions, he said, but has not yet hit mainstream use with voice teams at contact centers.
With contact centers and BPOs at pains to differentiate their service with customer satisfaction and customer intimacy metrics, AI remains a way from mimicking well the essential human component of the customer service industry. But with interest in technologies growing, it’s safe to say that questions around its reliability will be answered soon enough.