There is a lot of talk about leveraging AI for the Customer Experience (CX). This blog is intended to break down some of the key types of AI that can be used in contact centers for seamless customer experience interactions, and discuss some of the challenges and opportunities for the integration of AI.
To frame the discussion, here are a few steps a CX leader or program can take to get started with a Voice AI Strategy:
- Establish Frictionless and Reliable Access to CX Agents utilizing a combination of; Self Service Omni Channel Access, Wake Word Activation, Voice & Face Recognition & Authentication, and Speech-to-Text AI.
- Identify Current or Potential Platform Integrations; Leveraging Speech-to-text & Text and Voice based Analytics.
- Implement a Proof of Concept program: Cloud-based or On-premise?
Let’s dive in, and provide some context to these steps.
Establish Frictionless and Reliable Access to CX Agents utilizing a combination of; Self Service Omni Channel Access, Wake Word Activation, Voice & Face Recognition & Authentication, and Speech-to-Text AI.
We could start with customer onboarding because that’s where the customer experience really begins. Before diving into that, let’s consider the CX Agent side because it’s the CX agent that really shapes the customer experience. Even in the most automated of systems, there are, or there should be, a well-trained agent that the customer can access.
Access to Well Trained Agents. A well-trained agent is a critical backup that users can access when automation isn’t effective. Automation can fail for a variety of reasons ranging from ineffective technology to inquiries that are outside the domain of what was expected. A simple self-service technology for reaching an agent at any time can be a “wake word”. A wake word is a key phrase that is always being listened to by a speech recognizer. For example, an incoming caller could be greeted with a message like “Tell me how I can help you? If at any time you’d like to speak to a live agent, just say “Live Agent”.” This enables live agent access, but it is critical that customers gain access to a well-trained agent. There’s nothing more frustrating than talking to support staff that doesn’t know what they are doing, resulting in customers feeling their time is wasted.
One of the challenges today in working from home, with remote agents all over the world, is that qualified agents can get unqualified people to take their place. Even if they need to enter passwords to log in, the agent can provide the password to a friend. The solution here is “biometrics”. Biometrics can be face or voice authentication or a highly secure combination of both. The properly trained agent would only need to enroll once (it takes about 10-15 seconds) and then every time they log in their face and voice could be periodically and randomly checked in a completely unobtrusive way. This ensures that the customer is connecting with the well trained agent.
The Customer Experience with AI. The hope and expectation are that with proper use of great AI technology a business can get more happy customers, that churn less, and spend more. Onboarding is the first step in this process and the biometrics mentioned earlier can play an important role on the customer side as well as the agent. Biometrics can be enrolled; cards or IDs can be scanned, and the process of onboarding can be made faster and easier and result in a more secure biometrics-based system that is faster and more convenient to use.
Identify Current or Potential Platform Integrations; Leveraging Speech-to-text & Text and Voice based Analytics.
When a customer calls a contact center new types of AI can be deployed. Conversational agents sometimes called chatbots or domain-specific voice assistants can be used to carry on an intelligent dialog to help get consumers what they need faster. To make these work there needs to be accurate Speech-to-Text translation that translates the consumer’s inquiry into any language, an intelligent Natural Language Understanding model that can interpret what the user requested even in the presence of background noise, extraneous words, or phrasing errors, and a Text-to-Speech output to intelligently respond to the incoming caller in whatever language the caller needs.
Added value to the AI system can come from Text or Voice based Analytics that analyses the customer’s voice and words to better understand intents and emotions. For example, if frustration is building it might be good to get that live agent on without the keyword, or wake word, being spoken (or at least remind the consumer about how to access the Live Agent).
Implement a Proof of Concept Program: Cloud-based or On-premise?
A lot of companies including Sensory provide a lot of these technologies. But not all technologies and systems are created equally. Starting with a proof of concept program helps to clarify the needs and priorities for an individual organization.
For example, when it comes to systems or deployment of the AI solution, a CX organization may need an on-premise solution either for cost, security, or other reasons. Another CX organization may need a cloud-based, highly scalable bandwidth in a highly available system that automatically reroutes if any nodes on the system drop.
On the technology side, not all AI’s have the same capabilities or accuracy. For example, the wake word must work extremely reliably, if it accidentally goes off or doesn’t respond at a rate of just 2% the result could be thousands of frustrated customers. Biometrics are not all created equal either. Anti-spoofing technology to detect Liveness is needed so that a video or a mask can’t pass as the trained agent.
Choosing the right vendor or vendors (sometimes quality can be increased, and dependency decreased by using the best of breed in each category!) is not easy. Talk to Sensory and we can help you to understand the possible alternatives and help you customize your AI for the optimal Customer Experience.