June 8, 2026 | Henry C. Senturia
10 minute read
TL;DR: Implementation of Voice AI into customer service can help automate simple requests, but often increases customer frustration in high-emotion calls rather than helping. To responsibly implement Voice AI, companies should: 1) be transparent about their use of AI, 2) provide a clear route to a human agent, and 3) focus on building customer trust rather than purely optimizing AI agents for resolution speed.
I recently had the wrong package delivered to my apartment—it was addressed to my neighbor—but UPS notified me that the package I had ordered was successfully delivered, when it in fact had not been.
The “proof of delivery” photo they sent me proved only that they had delivered the wrong thing: a letter-sized package that was certainly not the large, 50+ pound package I’d been expecting. Realizing that I now had no idea where my package had ended up, I ruefully dialed the number for UPS customer service.
In my experience, customer service goes one of two ways:
I soon found myself trying to get the useless UPS AI agent (pretending to be a human) to just please connect me with a human person who would understand that I got the wrong delivery, firmly placing UPS customer service into category 2.
I had a much smoother experience with Economy Bookings when I needed to cancel a car rental and get a refund. After determining that I couldn’t solve the issue myself, I looked up the customer service number on their website. Right next to the phone numbers for each region, the site tells you that your call will first go to an AI agent to help route you to the right human agent, so I knew what to expect. To my surprise, I was actually able to cancel my rental and get a full refund just through the AI agent—the agent was trained well enough that I never needed to speak to a human. I trust Economy Bookings more and UPS less after these experiences—which is a common trend among customers. Zendesk found that 73% of consumers will switch to a competitor after multiple bad experiences, indicating that bad experiences lower a customer’s trust in a brand.
I’m the type of person who rarely calls customer support—when I do call, it’s usually for a relatively complex issue or request that I haven’t been able to solve myself by browsing the company’s website and using their automated self-service system. Voice AI is bad at handling complex issues that fall outside of its narrow training data, so when I do encounter AI agents, they’ve almost always been a roadblock that I try and breeze past as quickly as possible. Begone, bot.
However, voice AI can help improve a customer’s experience and reduce the number of calls human agents have to handle, so long as it is used responsibly. Automating simple customer service requests reduced customer experience (CX) resolution times by 69% for e-commerce clothing brand Shinesty. With that in mind, let’s dive into my 3 tips to make your business’ Voice AI a tool that helps customers rather than slowing them down or turning them away.
Transparency builds trust between CX teams and the customers they serve. Don’t pretend that AI agents are human. It’s weird! Misleading customers about AI use is a fast way to lose them. According to PWC, 32% of customers will leave a brand after just one bad experience. I and countless others feel betrayed when discovering that something we thought was human-made turns out to be AI-generated. This kind of betrayal shatters a customer’s hard-earned trust in your business because they feel like they’ve been tricked.
Eliminate these frustrations completely by stating up front that the customer is speaking to an AI agent. Responsible AI agent implementation already does this—Economy Booking’s website tells you outright that every call is first routed through an AI agent. In tandem with transparency, offer customers the choice to engage with AI—or to easily bypass it if they don’t want to engage with it—which leads us to the next tip.
A common frustration for customers interacting with AI agents is finding themselves stuck in a loop with a Voice AI that fails to understand their problem. Take, for example, an AT&T Voice AI that I spoke to, which would not connect me to a human agent when I asked to be transferred to one. The bot asked me to provide a reason for the transfer—but when I did, it directed me to read an FAQ page and then asked if I needed help with anything else, leaving me trapped in a loop. When I said “no” after a few cycles of this, the bot hung up. That’s not helpful. All a customer wants in a moment like that is to cut through the nonsense and reach someone that understands them. Make it easy for customers to do that. Per COPC, 99% of customers feel more comfortable with a human assisting to resolve their issues.
Voice AI doesn’t handle these high-emotion calls well—issues like delivery problems, disputes over billing, and requests for refunds—so ensure that customers always have a clear path to reaching a human agent.
COPC found that 88% of contact centers use AI-powered solutions as part of their CX operations, but, according to Qualtrics, only 29% of consumers globally trust organizations to use AI responsibly. It isn’t hard to see why. Heavy-handed corporate AI implementation that impedes customer or user experience has unfortunately become ubiquitous in the online world. PWC reports that 59% of consumers feel companies have lost touch with the human element of customer experience—myself among them.
Speed is important to customer satisfaction, but it can’t come at the expense of trust. One commonly-cited statistic used as an optimization target for AI agents is First Call Resolution (FCR), a measure of customer service success determined by calculating the percentage of customer service issues which are resolved by a single call. This statistic relies heavily on how a company measures it, which can lead to inaccurate representation of success. In a study by Customer Service Benchmarking Australia, an agent-reported FCR rate of 95% was revealed to actually be 67% when it was instead calculated based on post-interaction customer survey. The point is, optimizing purely for speed doesn’t always accurately translate into higher customer satisfaction.
Building consumer trust means ensuring that the implementation of AI in customer service is done responsibly. Think of AI as a tool to help your human agents work more effectively, not as a replacement for them. Using AI agents to automate simple CX requests is a good idea to help reduce the stress on human agents—people who are necessary for handling high-emotion calls where empathy and understanding are more important to a customer than speed. No matter how big or small your business is, you can set yourself apart by providing an excellent customer service experience that builds trust in your brand. I know that next time I need to rent a car, I’ll choose Economy Bookings, because I trust their customer service to smoothly handle any issues that I might come across. I can’t say the same about UPS.
Get in touch with us today to start brainstorming smart implementation for your Voice AI.