The Future of IVAs: How AI will Power and Re-shape Customer Engagement


Share on LinkedIn

Just over a decade ago, Apple released its iPhone 4S, “the most amazing iPhone yet.” It was faster, promised eight hours of 3G talk time, and introduced consumers to a feature called Siri, the first personal virtual assistant.

“For decades, technologists have teased us with this dream,” said Apple’s Phil Schiller, then the senior VP of marketing, at the 4S launch event in October 2011. “What we really want is to just talk to our devices, ask a simple question… and get a response.”

Much like the iPhone, AI-powered speech technology has evolved quite rapidly since 2011. Within the last five years, tech providers like Google and Amazon have launched their own cloud-based conversational AI services, making it easier for businesses to deploy the technology for customer service. Now, after decades of teasing customers with frustrating IVR experiences, brands are finally entering the first meaningful phase of being able to deliver on the dream of intelligent, voice-based self-service in the contact center.

Many contact centers are currently rolling out their first Intelligent Virtual Agents (IVAs), and it is exciting to think about how we will look back at this moment at the end of the next decade.

The New Generation

Within the next five years, I believe every major brand will have an AI-powered voice channel, in the same way that every brand has a website. Just like in the 1990s and early 2000s, when there was a rush by organizations to establish an online presence, businesses will be expected to have an intelligent, conversational interface to meet customers’ expectations for engagement.

Democratizing speech technology for all types of brands – not just those that can afford teams of data scientists and highly skilled developers – will be key to driving this adoption. We’re already seeing this happen with low-code / no-code, graphical user interface platforms that make it easy for non-technical professionals to build and manage their IVA workflows and integrations.

But effectively deploying IVAs that understand the specific vocabulary of your business, and the different ways customers phrase their questions and requests, requires ongoing training of your underlying conversational AI models. The treasure trove of data generated from voice and digital conversations in a contact center can be used to accurately train models. But this is currently a manual, time-consuming and expensive process.

In the very near future, leading IVA providers will solve this challenge by integrating more automated training methods into their solutions. And with this, organizations of any size – from SMB to very large enterprises – will be able to adopt conversational AI, unhindered by the labor-intensive process of building out the underlying algorithms. Businesses will be able to easily configure and monitor the performance of their AI models and derive meaningful insights from customer conversations in a cost effective and scalable way.

From there, we’ll see IVAs become capable of more context-aware, multi-turn conversations. For example, imagine a customer calls a restaurant to book a reservation. To complete this task, the IVA needs to confirm the date, time, and number of people in the party, but the customer may suddenly decide to ask a question about parking in the middle of that process. The well-trained, contextually aware IVA of the near future will be able to recognize the new intent, answer the parking question, and then pick up where it left off to finish confirming the reservation.

The Next Generation

As IVAs continue to evolve, they will become integral members of contact center teams, creating a “digital workforce” that provides pervasive assistance to customers and human service agents. They’ll automate increasingly sophisticated tasks and remain on the line (or in the chat) when interactions escalate to live agents. An IVA will be able to “listen” during an interaction, transcribe and summarize the conversation in real time, and intelligently surface helpful resources, coaching tips, or additional applications the human agent may need to resolve an issue. The IVA will even stay active after the interaction, logging notes and sharing insights to continually improve the agent’s performance.

In the slightly more distant future, I believe we’ll also see personal virtual agents that can make service requests on behalf of customers. Personal virtual agents could even become the next “killer app” for telecom companies. Just as voicemail and SMS were once transformative for the industry, every consumer may someday expect their smart device to come with a personal virtual agent that they customize to automate everyday tasks, such as scheduling a delivery or reserving a hotel room.

For example, a customer could simply ask their personal virtual agent to “book a dentist appointment for next week.” Like a trusted human assistant, the virtual agent would know which dentist’s office to reach out to, coordinate schedules with that office’s IVA, book the appointment, confirm the details, add it to the customer’s calendar, and set a reminder. And the customer wouldn’t have to lift a finger to make any of this happen.

For now, this type of use case is purely hypothetical, but IVAs are already enhancing customer engagement in ways that didn’t seem possible just a few years ago. The technology is more life-like and ubiquitous than ever, but “What’s next” is no simple question. It will be up to businesses and their customers to provide the response. Afterall, there is a difference between what’s possible and what is useful.

Callan Schebella
Based in San Francisco, Callan Schebella is EVP of Product Management at Five9. Callan has worked his entire career in Voice Automation, Machine Learning and related technologies, with a focus on creating solutions that make it easier for businesses of all sizes to build and maintain speech-enabled applications.


Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here