IVAs: Using AI to Serve Customers and Contact Centers
Intelligent virtual agents (IVAs) represent the future of omnichannel self-service, a new standard of voice and digital self-service in a channel-optimized format. They can also provide contact centers with context-aware guided support and relevant information for each individual customer interaction, for agent-assisted or escalated self-service interactions. And with the COVID-19 pandemic putting increased stress on businesses, customer service departments, and remote workers in need of support, these are welcome developments (see related story on page 24).
DMG Consulting defines an IVA this way:
A system that utilizes artificial intelligence, machine learning, advanced speech technologies (including natural language understanding/natural language processing/natural language generation [NLU/NLP/NLG]) to simulate live and unstructured cognitive conversations for voice, text, and digital interactions via a digital persona.
IVAs—which are powered by sophisticated underlying technologies that enable them to simulate cognitive reasoning and respond to human beings in a conversational manner in phone and digital interactions—support omnichannel environments so customers can start in one channel and move seamlessly to another. IVAs leverage machine-readable, context-aware knowledge bases (or other data sources and repositories) to store and retrieve the data needed to respond in a personalized and contextually relevant manner to human questions or input. They can mimic human cognitive functions such as understanding and sentiment, and they use machine learning to continuously improve their accuracy and effectiveness over time; their intelligence is continually evolving based on knowledge gained from each interaction, which is assimilated and leveraged in future conversations and transactions. IVAs may include visual representations—i.e., avatars—on websites or within mobile apps, or they can just be speech-enabled.
There are dozens of artificial intelligence (AI) technologies available today, but the three that are core for IVAs are NLP/NLU/NLG, real-time analytics, and machine learning (ML).
The NLP grouping of technologies—which include transcription, speech-to-text, and text-to-speech—allow organizations to understand what customers are saying; they’re designed to find meaning and insights in conversations, whether spoken or written. NLP enables computers to understand the meaning without a predefined syntax for the content. It also allows computers to respond to people in their own language. Practical applications of these technologies in contact centers include speech analytics, IVAs, robotics, and other capabilities.
Real-time analytics encompasses a highly diverse group of technologies and applications. Real-time analytics frequently takes and acts upon the input from an NLP solution. It may also draw upon historical data, a customer relationship management (CRM) solution, a sales system, marketing databases, inventories, etc., to determine the most appropriate action to take. The challenge is to do this fast enough to enable a live or automated “agent” to act in near real time. Billions of dollars have been invested in a large variety of real-time analytics solutions over the past 20 years. All of the older applications were rules-based, and most were too slow to be helpful in a real-time servicing scenario, which is what is needed in a contact center.
Machine learning, an essential component of AI, is a highly complex set of algorithms that can automatically “learn” and identify trends, patterns, opportunities, etc., in a dataset. (A dataset may include recordings or transcribed texts, such as tweets and emails.) ML can operate in three modes: supervised, semi-supervised, and unsupervised. It can be used to provide the most current information about the customer journey, which should be a priority for all service organizations. Machine learning is already being used in a growing number of contact center solutions, including speech analytics and workforce management.
Additionally, in the era of AI and real-time analytics, data repositories are a must-have. AI-enabled solutions enhance self-service and agent-assisted interactions by accessing knowledge bases, other data repositories, and third-party sources to locate and retrieve the most appropriate resource material to provide to a prospect or customer, or to answer questions.
IVAs are the future of omnichannel self-service and for helping employees, not just contact center agents, perform their jobs. In the past couple of years, billions of dollars in research and development have been invested in AI-related products, and IVAs have been the beneficiaries of a lot of this investment. There’s still a lot more work to be done before IVAs can come close to replicating an interaction between a live agent and a customer, but some of the IVAs are already significantly more effective, responsive, flexible, and accurate than IVRs.
Donna Fluss (email@example.com) is president of DMG Consulting, a provider of contact center, analytics, and back-office market research and consulting.