The slightly pesky, goggle-eyed paperclip affectionately known as Clippy that hovered in the right-hand corner of Microsoft Word was a precursor to the sophisticated desktop virtual assistants of today.
While Clippy was an overeager but helpful medium for navigating Microsoft Word’s labyrinthine knowledge base, today’s virtual assistants provide user-friendly front-end interfaces that are capable of performing and automating back-end tasks, combining human-facing artificial intelligence and natural language processing with rote robotic process automation.
“The greatest innovation the robotic process automation market has yet to see.”
The NICE virtual assistant, known as NEVA (NICE Employee Virtual Attendant), was developed with frontline customer support agents in mind - namely, eliminating the hellish parts of their job. According to research by NICE, agents spend 25 percent of their time on repetitive tasks like copy-pasting data from one CRM system into another, trawling knowledge bases for key information and filling out forms - sometimes handling all three while on the phone with a customer.
These thankless routines “don’t really leverage their skills and training,” Karen Inbar, solution marketing manager at NICE, said in a recent webinar with CCW Digital, ‘Case Study: The First Employee Virtual Attendant.’ Given recent strides in self-service tools, customers tend to call live agents with complex queries, making immediacy of information access all the more important. The NEVA virtual assistant provides real-time guidance and compliance scripts by monitoring the agent’s desktop activity.
At the keynote for the NEVA launch in Orlando, Oded Karev, VP of Robotics Process Automation at NICE, introduced the product as “the greatest innovation the robotic process automation market has yet to see,” before whipping a black cloth off a concealed flat screen TV with a Jobsian flourish.
“I’m smart, really smart,” NEVA, who vaguely resembles a LEGO character, chirped from the screen. “Some people say I have artificial intelligence, but I’m not insulted by that.”
In a session of show-and-tell, Karev and NEVA demonstrated an inbound customer call to a hypothetical cable company, where the customer wanted to upgrade to a full sports channel package. NEVA, listening to the conversation, pulled up a series of callouts, first identifying that the customer was a premium member who is eligible for an upgrade for the same price.
When the delighted customer says yes, he’ll take it, NEVA automatically signs him up on the backend by auto-populating a series of forms, adds the additional channels to his account, and then generates an email summarizing the call that is automatically sent to the customer. NEVA even turned her head to follow Karev as he walked from one side of the stage to the other.
The company insists that the technology is designed not to replace human agents but to help them “reach their fullest potential” by squaring away the so-called “boring” parts of their job. Thought leaders in this space strongly advocate for AI as an antidote to repetitive job functions, but watching NEVA in action makes it hard not to feel that this wisecracking machine with a cute button nose has potential way beyond that.
NEVA doesn’t wait patiently to be called upon; she can pop up in response to a keyboard stroke, mouse click, or when the agent picks up the phone or loads a specific application. She emerges with a corresponding callout containing links to relevant information, compliance scripts, or checkboxes for suggested next steps, depending on the situation. If the agent starts a sales process, NEVA can trigger a relevant sales script, and then automate the sales process by closing the sale “behind the scenes and freeing the agent for the next call.”
“If it’s a complaint they’re handling, then NEVA can suggest some retention offers for that customer,” Inbar explains. “If there are any compliance-related scripts that the agent needs to read, NEVA can pop those up at the right place and the right time.”
NEVA also consolidates all the information an agent needs for a specific call, including relevant links and customer data such as recent purchases, complaints or pending payments. “Instead of the agent wasting time looking for that info, NEVA will bring everything they need in the context of the topic of the call to a single screen displayed to that agent,” Inbar elaborates.
As with any other virtual assistant like Siri or Amazon Alexa, agents can initiate conversations with NEVA through voice or chat to ask her to do specific things. For example, a financial services company might receive a call from a customer asking to up their credit limit. The agent can ask NEVA to file a request on the back-end by enabling the agent to specify the amount of the credit limit increase, the reason for it, and any other data needed. Then, NEVA triggers a backend robot to file the request.
With so much of the back-end process being automated - in some cases with or without the push of a button - one can’t help but wonder how this affects agent engagement. If transactional tasks are handled by robotic process automation and even higher-level functions like upselling and cross-selling are facilitated by scripted suggestions, what, then, is left to the agent? Contact centers that use such advanced technologies have to rejigger agent workflows or even job descriptions to accommodate “higher-level” work, else staff are left to twiddle their thumbs.
Many companies espouse agent empowerment - giving agents leeway to go off script and exercise some discretion over compensating an unhappy customer outside of written policies - but when provided with a prescient tool that does most of the work, agents can become dependent on it.
This raises questions not only for agent engagement but their future job mobility - what happens when the agent applies for a job at a different contact center that doesn’t use technology like NEVA, where they have to do actual work and think for themselves?
At the risk of sounding like a curmudgeonly technophobe, NEVA seems to offer a level of automation that’s excessive at best, counterproductive at worst.