Making Self-Service More Intelligent
View this article on the publisher’s website
Omnichannel self-service solutions are a requirement for companies that want to deliver a cost-effective and consistently outstanding customer experience (CX). Intelligent virtual agents (IVAs)—a.k.a. chatbots, voicebots, or just plain bots—can automate many tasks and inquiries previously handled by live agents, reducing operating expenses.
But cost reduction is only one reason companies need intelligent self-service solutions; consumers are showing an increasing preference to help themselves. Additionally, as the volume of contact center voice and digital interactions continues to rise, organizations are finding it difficult to identify and hire enough qualified people to staff their service departments. The way to address each of these needs is by staffing contact centers and CX functions with an optimal mix of IVAs and live agents to handle their unique volume and types of interactions. The rapid progress of artificial intelligence, especially generative AI, is leading to vastly smarter and more capable bots.
Leveling Up Bots
Intelligent self-service applications are based on several AI technologies, including machine learning, advanced speech technologies (e.g., natural language processing (NLP), natural language understanding (NLU), and natural language generation (NLG)), deep neural networks, and predictive analytics. These underlying capabilities enable IVAs/bots to simulate live, unstructured conversations in voice- and text-based channels, identify customer intent/sentiment/emotions, automate completion of customers’ requests, and escalate the interactions they don’t yet know how to handle to human counterparts. As sophisticated as these conversational self-service solutions have become in recent years, the application of generative AI is taking them to an entirely new level.
While the first generation of AI-enabled self-service bots successfully automated many inquiries, when confronted with undefined intents, the bots’ only option was to transfer the person to a live agent to complete the interaction, even though the customer wanted to self-serve. Generative AI greatly expands conversational understanding of unstructured input (i.e., questions) in these self-service solutions via a large language model (LLM), enabling them to identify an appropriate answer. Generative AI is also improving and speeding up self-service solution testing, making the solutions better from the outset. Although there is a tremendous amount of work yet to be done, generative AI has already increased bots’ operational capabilities. It enhances their language understanding, improves response accuracy, and can deliver comprehensive summaries of self-service interactions to a customer relationship management (CRM) solution or other servicing system so live agents know what has been done for a customer.
Ease of Deployment Is a Must
Today’s IVA solutions come with low-code/no-code development environments that allow nontechnical business users to build and implement them via graphical user interfaces (GUIs) rather than programming environments. These GUIs contain the elements, objects, and application programming interfaces (APIs) that let them integrate with CRM solutions, knowledge bases, and other operating systems; and the communication components needed to create self-service workflows and automations. In many IVA initiatives, the vendor or a systems integrator builds the initial solution and then hands it off to the end user to maintain after training.
Today’s IVAs are intended to be omnichannel solutions that allow companies to build once and deploy in multiple channels, like voice, email, chat, SMS, WhatsApp, and more. This reduces costs and enables customers to receive the same answer regardless of their interaction channel. As marketing and contact centers work more closely together, IVA functionality is starting to be deployed via the company website, facilitating more standard responses and improving the CX.
Final Thoughts
Companies and consumers alike appreciate the necessity and benefits of omnichannel conversational AI-based self-service solutions that can accurately respond to questions or complete requests. The technology needed to support these capabilities more fully is just becoming available, although the market has been close to getting there for the past few years. Issues surrounding delivery of an unlimited, truly conversational AI-based self-service solution are not simple or inexpensive. These technologies are new, best practices have not yet emerged, and companies must make trade-offs between system flexibility, relevance, and accuracy to avoid bias and hallucinations. Given the current and expected future state of the technology, as well as the market’s interest and readiness, DMG expects to see major progress in self-service solutions during the next three to five years. The question every organization needs to ask is how long it can wait to enhance its self-service environment, as there is a major opportunity cost with each passing day.