March 19, 2018
By Paul Korzeniowski technology writer
Features

Intelligent Virtual Assistants: Gaining Traction but Confronting Challenges in the Enterprise

Voice is becoming a common interface in the workplace as well as in the home. Everyone is quite familiar with Apple’s Siri, Microsoft’s Cortana, Amazon’s Alexa, Google Assistant, Samsung’s Bixby, and the host of similar solutions for consumer uses. Business applications haven’t been as prevalent, but with speech recognition technology improving and data processing power growing, it’s no surprise that so many corporations are developing intelligent virtual assistants (IVAs) to help customers and employees speed up tedious tasks.

“Large enterprises realize that customers and employees can do more via voice interfaces,” says Dan Miller, founder and lead analyst at Opus Research. “They see a clear path from their investments to tangible business benefits. The technology was big in 2017 and will be even bigger in the coming years.” In fact, research firm Tractica predicts that enterprise virtual digital assistant users will top 1 billion by 2025.

IVAs are automated self-service resources. The technology relies on natural language processing, machine learning, semantical search, conversational analytics, and knowledge management to present information to users. Typically, customers and employees enter information, often in the form of speech. The applications recognize the input and provide helpful responses, ones that move a repetitive process one or more steps closer to completion.

These products have great promise. In some cases, the solutions are smart enough to rapidly recognize intent. They can categorize the purpose of interactions by examining recent activity, location, preferences, and entitlements. The ultimate design goal is to have the system respond with the most relevant information or appropriate action.

A few recent technical advances have helped the technology move out of the lab and into the workplace. A couple of years ago, speech recognition rates were circumspect, but vendors have improved their systems so that now about 90 percent of all voice interactions are correctly understood.

Also, businesses have been building up their knowledge bases and making them more accessible to customers, employees, and partners. In recent years, firms created digital product literature, customer relationship management records, chat transcripts, and Frequently Asked Questions files. Rather than being stored in employees’ heads or on top of their desks, these records have been formatted so they are searchable by customers, employees, and partners.

Consequently, with IVAs, companies offer ways to service customers more efficiently and less expensively. In fact, Opus Research projects that IVA revenue will increase from $1 billion globally in 2016 to $4.5 billion by 2021.

A Widening Reach

Potential business uses have also been growing exponentially, and that is not expected to let up anytime soon. In many cases, the initial focus was on customer service and the contact center, a niche that continues to be of keen interest. Horizontal and vertical uses of the technology have been emerging as well. Take, for example, the voice-enabled Alexa for Business solution implemented by Amazon in November. It’s geared to helping employees manage tasks such as scheduling appointments, sifting through messages, starting meetings, controlling conference room equipment, and updating the status of pending sales.

In addition, vertical market solutions are emerging. “IVAs can automate many repetitive business processes,” Miller says.

Financial services companies can conduct conversations with mortgage customers rather than having them fill out complex applications. In the travel industry, assistants can help travelers make airline, hotel, and rental car reservations. In government, these services can help people pay their property taxes or take complaints. Amazon is pushing the technology into the retail sector and enabling consumers to place orders by talking rather than entering text.

But the overall IVA market faces a few challenges. First, the term “IVA” is nebulous at best. “When companies talk about virtual assistants, they include a range of technologies: bots, software relying on natural language processing, and sophisticated avatars that hear, respond, and recognize emotions,” Miller says, noting that hundreds of thousands of bots exist but fewer than 100 of them could really be called IVAs.

Not surprisingly, market competition has also been growing. Opus determined that the number of vendors offering enterprise intelligent virtual assistants more than doubled last year. Some vendors developing such solutions include Amazon, Aspect Software, Interactions, IBM, IPsoft, MindMeld (acquired by Cisco Systems in May), Next IT (acquired by Verint Systems in December), Nuance Communications, Omilia, Samsung, and [24]7.ai.

Nuance sells Nina, a system that includes natural language capabilities, machine learning, a conversational experience, and analytics. [24]7 focuses on customer queries and self-service answering. IPsoft markets Amelia, an assistant designed to be aware of the full context of every conversation so that she can adapt her social tone and actions accordingly.

Omilia, whose solutions are geared at large companies, has a solution that has handled more than 1 billion interactions, supports 16 languages, and has been deployed in about a dozen countries.

“Omilia was one of the few vendors who refrained from joining the marketing craze aimed at grabbing the AI buzzwords and mindshare,” says Dimitris Vassos, its CEO. “Instead, we focused on growing a solution built with actual enterprise customer service in mind.”

The hype around AI, Vassos adds, actually hurt the industry as a whole. “[Last year] was the year conversational technologies were supposed to go mainstream. Tech giants and start-ups alike came out promising the world the holy grail of artificial intelligence with human-level conversational performance. Just like that, the chatbot era began, sped up the runway, and crashed before it even flew its first mile. Most solutions were designed in isolation of the technical, legal, and business challenges that are present in real enterprise contact center environments. These solutions proved to be far from ready for prime-time enterprise production roll-out,” he states.