Jacada Visual IVR advertises that it plugs “into your existing Avaya, Genesys or Cisco IVR by reusing your VXML.” This allows now current deployment and a fast win.  But it has only swapped an IVR script which is not functioning agilely with an interaction panel more palatable to the customer market. 

AI-driven call agent technology exists today. VPI VirtualSource is a hosted solution that leverages conversational virtual agents powered by Artificial Intelligence to effectively parse and automate a variety of inbound and outbound calls. AI virtual agents provide a set of responses and skills that mimic the responses of human agents.  With proper back end system integrations a wide and open ended set of responses is possible.  They even provide for proactive responses by linking past associated information requests to the current interaction flow. It is integrated with Avaya, Nortel, Cisco, NEC, and Siemens.

Both these products still leverage existing IVR technology. While they may bounce through the cloud, they are only active during a user initiated IVR session. We can go further by piecing together existing technology developed outside of ICE.

Intelligent Voice Agents

Consumers today have access to a multitude of intelligent agents which have integrated cloud-based natural language systems. Amazon’s Alexa via Echo & Fire is flowing into households.  Apple’s Siri answers the queries of iPhone users accessing website data after recognizing speech via Apple partner Nuance’s Natural Language Recognition engine. Google Now/Google Assistant is in Google Home, used in search, Android phones, present in Chrome, and is migrating out to many interactions with Google services. Microsoft embedded Cortana into Windows 10. Facebook is fielding M. Samsung has S Voice. HTC has Hidi. LG has Voicemate. These applications are far from perfect.  Users collect the hilarious inaccuracies that sometimes occur.  But seriously, between speech recognition and then integration with the responding application, these agents are still early technology.  But enormous investment and fast development is occurring – much faster than in IVR systems.

Strong privacy concerns about these agents exist. If the application is on, it is listening to its environment. Amazon assures that the Echo units only stream recordings from the user's home to the Alexa agent when the 'wake word' activates the device. But the Echo device is always listening for that word. According to Slash Gear, Amazon streams digital recordings of user’s speech audio  just before and after the ‘wake up word’. These recordings are used to improve speech recognition. Amazon’s official privacy policy does not address how long these audio streams are stored. In an ironic twist, Amazon may have a response to these concerns.    A query to my Echo, “Alexa are you listening to everything we say?” Results in the response, “Shuffling Wedge’s music library.” Classic misdirection with good music. 

It may be that every cloud-based virtual assistant, including Google, Siri, Cortana, is collecting every voice interaction, every keystroke, and every menu selection as it seeks to be ever ready to assist. A more current concern is surfacing about Microsoft’s Windows 10 Cortana because it now cannot be turned off. Still users must intentionally enable apps and services for use by Cortana, Google Now, and Alexa. But the more you enable, the more it can do. The more you enable, the better they learn your habits and actions, enabling better predictive assistance. The easier your life becomes.

If traces of a user’s interactions are stored in the cloud, or retrievable from a user’s personal device, it is subject to search and seizure. Not just subpoena driven search and seizure. In selected instances, international governments, business competitors, and criminal hackers prize this data. This is more than a privacy concern. As more personal and business activity occurs in the presence of these agents, the potential to reverse engineer human interactions becomes likely. This includes reverse engineering and manipulating automated processes. This is the dark side of the future.