When properly designed and built, AI assistants can have a great impact. Companies have been designing chatbots and voicebots long enough to know their advantages and disadvantages. While both are great at resolving simple and high-frequency use cases (although you should aim to do more eventually), chat and voice channels aren’t always the most optimum channel for all use cases. There’s a better alternative that’s far easier, cheaper and quicker than automating every use case in every channel from end to end. It’s an approach that United Airlines has taken to create an award-winning CX solution.
This article explores that solution, as well as the limitations of single-channel AI automation. Then, it looks at the potential of multi-channel integration. It summarises many of the points made in a recent webinar: ‘The hidden ceiling in CX automation and how to shatter it’ featuring Andrei Papancea, CEO and Co-founder, NLX along with VUX World’s Kane Simms.
Channels aren’t interchangeable
Many businesses rely on AI-powered automation to improve customer service, but using only voice or chat can often lead to problems.
Voice-only AI struggles with complex tasks. If you’ve ever tried spelling your name or resetting a password over the phone, you know how frustrating it can be. Speech recognition systems often misinterpret names, numbers, and special characters, making these processes unreliable. Over voice, a simple spelling mistake can lead to failed authentication.
Voice interactions also require customers to remember a lot of information. For example, if an AI assistant explains a refund policy over the phone, the user might forget key details and need it repeated. Some processes, like booking travel, are easier with a visual interface where users can quickly compare options. Customers may need to confirm details, make selections, or change their minds mid-conversation, which can lead to errors or inefficiencies.
Chatbots have limitations too. While they work well for simple questions, they can be slow and frustrating for multi-step processes, like applying for a mortgage or changing a flight.
Another challenge is that AI often lacks context. Customers may start with a chatbot and then switch to a phone call, only to find they have to repeat everything.
Plus, many businesses already have well-functioning websites and mobile apps for self-service. Instead of creating new chat-based workflows, businesses could better guide users to existing digital tools – something many chatbots fail to do.
AI support for those who need it
Voice+ is designed to solve many of the challenges faced by AI-only automation by combining voice guidance with digital interfaces. Instead of relying solely on voice or chat, it helps users complete tasks while interacting with a website or mobile app.
Here’s one example of how it can work: when a customer calls a business for help, the AI assistant sends them a secure link via SMS or email. This link opens the relevant self-service page, and the assistant stays on the line to provide step-by-step guidance as the user navigates the visual interface. Errors are reduced because the user can ask the AI assistant for help while they progress.
The system can work in two ways:
- Unidirectional: providing instructions without requiring the user to respond.
- Bidirectional: where users can speak back to the AI for more interactive guidance.
By integrating voice and digital channels, Voice+ removes the need for users to memorise complex steps or repeat information. It also allows businesses to get more value from their existing digital tools instead of forcing every interaction through AI-only automation. If needed, the AI can escalate to a human agent while keeping full context, ensuring a seamless transition. Customers don’t always want AI to complete a task for them – they might just need guidance on how to do it themselves.
How United Airlines improved accessibility requests
Let’s look at how a multimodal approach improved United Airlines’ user experience.
United Airlines faced a challenge in handling accessibility requests, such as adding wheelchairs and service animals to flights. These requests are complex, requiring passengers to provide details like wheelchair type, battery specifications, and other important information. Traditional voice-only automation struggled with these tasks because it required precise input and multiple decision points.
To improve the process, United Airlines added Voice+, allowing passengers to call and receive a secure link to a self-service form. While the user filled out the form, an AI assistant provided real-time guidance, helping them complete each step correctly. This ensured that important details were entered accurately, reducing errors and confusion. If a passenger needed additional help, the AI could escalate the request to a human agent while keeping all relevant context, preventing the need to repeat information.
The results were significant. 64% of these requests were fully automated, reducing the workload on customer service teams. At the same time, customer satisfaction reached 90%, showing that users found the process smooth and effective. United’s success with Voice+ earned them a 2024 Opus Research Conversational AI Award, demonstrating how AI-powered guidance can simplify complex, multi-step interactions while improving efficiency and user experience.
Don’t replace what already works
One of the biggest challenges in AI automation is the need for complex integrations when creating new workflows. Many businesses already have well-functioning self-service options on their websites and mobile apps, but traditional AI solutions often require rebuilding these processes from scratch.
You could take a different approach, though. Enhance your existing digital tools instead of replacing them. You don’t need to force users to complete tasks entirely through voice or chat. It’s possible to guide them through a digital interface with a voice assistant. This means companies don’t have to redesign their workflows or build new AI-driven expeariences for everything.
This system can be quick to deploy. Just add a lightweight synchronisation SDK to your website or app, which is similar to setting up analytics tracking. This eliminates the need for deep backend integrations while still providing seamless AI-driven assistance.
For customers, this makes self-service easier and quicker. They get real-time guidance while interacting with familiar digital platforms, reducing frustration from voice recognition errors or slow chatbot exchanges. Over time, this could also increase self-service adoption, lowering call centre volumes while improving overall efficiency. If extra help is needed, the AI can pass the case to a live agent with full context, ensuring a smooth experience.
Learn more
You can learn more about Voice+ here, and if you’d like to see the full deep dive into this working in practice, you can watch the full webinar here.
