Virtual agents are changing the customer experience. As advancements in artificial intelligence (AI) technologies enable virtual agents to perform transactions that were once only addressable by a team of live customer service agents, financial services, healthcare and other industries are realizing the dramatic impact virtual agents can have on contact center operations and customer interactions. In fact, according to PwC research, 27% of consumers weren’t sure if their last customer service interaction was with a human or a bot. And 62% of consumers believe that AI can reduce the time it takes to get answers while still being highly tailored to their preferences.¹ In addition to enabling customers to perform common tasks and get answers to routine questions, the rise in virtual agent as customer service agent brings several benefits to business, including:

• 24/7 availability
• Lower labor costs
• Ability to handle multiple chats simultaneously

How virtual customer agents are different from interactive voice response

Unlike interactive voice response (IVR), which comprehends speech within a defined scope of programmed keywords, virtual customer agents leverage natural language processing to create an experience that is more human sounding and conversational. Because IVR results in a voice command system that is only as good as its application in limited situations, it often leads to frustration as customers repeat words that the system doesn’t understand, resulting in them giving up on the transaction or seeking help from a live agent. Virtual agents, on the other hand, facilitate an end-to-end customer experience by delivering information, taking action on behalf of the customer, performing transactions and engaging in a manner that is more like that of a live agent. Let’s look at the questions we hear most frequently from organizations considering adding virtual agents to their contact centers.

Virtual customer agents: Answers to your top five questions

Question #1: What happens if the virtual agent doesn’t understand the customer? Answer: When a virtual agent doesn’t understand what the customer is saying, there are several options to resolve this situation. First, the agent can ask the question in a different way. For instance, if a customer calls with questions about the benefits of an insurance policy, the agent can ask clarifying questions to help get to the core issue. In this way, the agent can scope the issue so that it can get to the right answer. Another option is for the virtual agent to simply acknowledge that it is not understanding what the customer is asking for, and then offer to send the customer to a live agent or to an online resource that may be able to help. Finally, an organization can set expectations upfront by letting customers know what a virtual agent can and can’t do, for example, “I am designed to help you set up your checking account. If you need help with something else, I can connect you with a different virtual agent.”   Question #2: What happens if background noise or interruptions make it difficult for the virtual agent to hear the customer? Answer: Much like any situation with background noise that makes it difficult to discern what a customer is saying, there is a certain level of cacophony that a virtual agent can parse. While one live agent may be able to use context to hear through confusion better than another, there will always be situations that just make it too difficult for a person or virtual agent to make any sense of it. A virtual agent does its best to get the words from the audio signal to process and understand, but in situations where it can’t do that, the virtual agent can use signal processing to make as much sense as possible out of the signals that do come through. When that still doesn’t work, the virtual agent can take the same actions a live agent would take in this situation by letting the customer know that he or she is having difficulty ascertaining what’s being said, and then offering alternative suggestions to get the help he or she needs.   Question #3: What’s the best way to handle the disclosure that a customer is talking with a virtual agent? Answer: Customers calling into a contact center are looking for help; ultimately, they want to successfully resolve the reason for their call. Most companies believe that the best way to help people and make them feel understood is to be upfront about the system that is helping them. Gridspace encourages the organizations using our virtual agent solutions to let customers know the capabilities of the system and how it can help them, for instance, “Hi, this is a virtual agent, and I can help you with answers to questions about your policy.” We’ve found that customers react positively to this because they know that they can get their questions answered or problems solved. In contact centers that are exploring the use of live agents and virtual agents interchangeably, the goal is to make the hand-off as efficient and seamless as possible by letting the customer know that the live agent will help to address the reason for their call.   Question #4: Do virtual agents learn? Answer: Virtual agents turn an audio signal into language; then, by applying meaning and understanding to that language, generate language back to the customer—this is what creates a conversation. Many parts of that process benefit from learning. Gridspace looks at how customers talk to human agents and then leverages that insight to help virtual agents interact with, and respond to, customers. It also helps to identify when a customer says something important and ensures that virtual agents don’t miss these cues. Because Gridspace builds our systems with models that understand conversational speech, these models can find patterns that contribute to continuous improvement in virtual agent performance.   Question #5: How can virtual agents be rolled out? Answer: Gridspace has worked hard to make the rollout of virtual agents easy. In fact, with Gridspace, contact centers can spin up virtual agents in just a few minutes. This is possible because we make both the telephony system and the neural network that is able to access the audio with the caller. By reducing the process to just a few steps—creating the phone call, determining what the virtual agent will do and identifying what will happen with the information coming out of the call—rolling out a virtual agent, and realizing the benefits of this additional contact center operation, is fast and easy. Explore how Gridspace can help you deploy virtual agents in your contact center.

• 1. PwC, Bot.Me: A revolutionary partnership How AI is pushing man and machine closer together, 2017.