Often compounding the problem for some of the “tortoise technologies” are their inherent complications. Two that come to mind are the VUIs (voice user interfaces) and the IVAs (intelligent virtual assistants). Both require systems that can imitate human capabilities—speech recognition and decision making.

The VUI made its appearance in 1975 when Dr. James Baker and his wife, Janet, presented a speech recognition system for PCs called Dragon Systems. Today, Dragon Dictation is a popular app both on Windows 10 and Apple’s iOS, and there are several more powerful Dragon Naturally­Speaking programs available. The development curve for these natural interfaces has traced a long, patient arc over the decades.

IVAs have also cautiously turned a corner. The polite, feminine voices of intelligent virtual assistants can now be heard on tablets and smartphones. Apple’s Siri and Microsoft’s Cortana have recently been joined by Amazon’s Echo—the black cylinder on the coffee table that will talk to you as it programs your next set of playbacks from your music library.

Just last month, the team that developed Siri announced a new kind of IVA called Vivian, an AI (artificial intelligence) personal digital assistant. It uses something called “dynamic program generation” software that can write a program in microseconds in order to answer questions like, “Hey Viv, will it be warmer than 70 degrees near the Golden Gate Bridge after 5 p.m. the day after tomorrow?” Dag Kittlaus, cocreator of Siri and now the CEO of the company developing Viv, says the app will be rolled out in stages later this year.

As you might imagine, the grafting of a natural speech interface onto an intelligent computing engine that can reprogram itself, midquestion, would take quite a bit longer to wire together than many other innovations.


After 41 years of slow progress, speech-enabled IVAs that can converse both with customers and employees are now in development. Companies like Capital One, Domino’s Pizza, Hyatt Hotels, and Shell Oil are just a handful of those incorporating the technology.

One of the simpler tasks assigned to enterprise virtual assistants is to answer customers’ frequently asked questions. They do this without dragging them through the branches of a decision tree, which could sound something like: “If you answered yes to the previous question, press the pound key, skip questions 3 and 4, and scroll down to question 5.”

Dan Miller, lead analyst and founder of Opus Research explains, “Enterprise intelligent assistants do a better job of understanding what you are asking and then responding to it. They take turns, so it becomes conversational speech.”

Rebecca Jonsson, chief researcher for Artificial Solutions, sums up the progress this way: “We have gone from virtual assistants that could help out navigating Web pages (such as IKEA’s Anna, fluent in 20 languages) or virtual assistants that helped out in customer service (answering frequently asked questions), to intelligent virtual assistants that can perform tasks, take over well-defined processes, and solve issues for the enterprise by becoming digital employees.”

Cited in Speech Technology, Jonsson explains an additional advantage of IVAs: “The interaction with the consumers becomes independent of the device, operating system, and consumer’s technology expertise.” There’s no need to learn structured commands or enterprise technology.


Kasisto is a personal assistant for finance and commerce that can be embedded in mobile apps. Used by banks, wealth management, and credit-card companies, Kasisto can understand conversational voice and text commands and can answer more than 1,000 banking-related questions. It uses natural language processing, AI reasoning, and machine learning to understand ­questions and create responses. (See www.kasisto.com.)

DigitalGenius has an AI customer-­service automation platform that’s available to large company call centers. It uses deep learning to master a customer’s ­service-related chat logs, e-mail transcripts, and Facebook and Twitter messages to create algorithms that produce customer-service software. The system is available as a monthly rental. (See www.digitalgenius.com.)

When Dag Kittlaus’s Vivian is released later this year, the app may represent a significant stride forward for these developments, but it will also be just another step in the long journey of aspiring virtual personal assistants.

About the Authors