Customers Are Not Rejecting Voice AI. They Are Rejecting Generic AI on the Phone.

A customer notices her check engine light on the drive to work. Before she calls the dealership, she does what most customers now do: she checks the website, tries the online scheduler, and finds nothing open that fits her week. At 8:40 on a Tuesday morning she calls the service department. An AI voice agent answers, asks her to describe her issue, and then tells her she can book online. She explains that she already tried. It asks her to rephrase. She says “representative” twice, gets transferred, and lands in a voicemail box because both advisors at the drive are writing up walk-ins. By lunch she has booked with the independent shop down the road, and nobody at the dealership knows she called.

That customer is not frustrated by voice AI as a concept. She is frustrated because the system that answered had no idea what a service appointment actually involves, could not see the schedule, and had no way to hand her to a person with any of the context she had already given. That distinction matters, because a growing share of the conversation about AI in customer service is treating the two as the same thing.

Why the Backlash Against AI Customer Service Is Real

The frustration is well documented. A Gartner survey of more than 5,700 customers found that 64 percent would prefer companies did not use AI in customer service, and the most common concern was that it would make reaching a person harder. The same research found that 53 percent would consider switching to a competitor if they learned a company was going to use AI for service. Forrester’s 2025 CX Index recorded a fourth consecutive year of decline in customer experience quality, pointing in part to “disappointing implementations of potentially game-changing technology (including AI).” Even Klarna, one of the most publicized adopters of AI support, reversed course and began investing in human agents again, with its CEO stressing that customers need to know “there will always be a human if you want.” Gartner now predicts that half of the organizations planning significant customer service workforce cuts will abandon those plans by 2027.

A recent LinkedIn discussion started by a customer experience leader venting about “horrible AI gatekeepers” drew dozens of responses from CX executives, contact center operators, and AI builders. What stood out was how consistent the diagnosis was. Very few people argued that the underlying technology is incapable. Almost everyone described the same failure: AI deployed as a cheaper front door, with no access to the systems behind it, no sensible path to a person, and success measured by how many calls never reached a human rather than how many problems were actually solved. One commenter summarized it well, noting that most bad experiences come from three gaps: no access to real systems, no clear handoff, and no one reviewing real conversations after go-live. Another observed that most voice agents on the market have simply made the phone tree more talkative.

Why Generic AI Struggles on a Dealership Phone Line

For a dealership, these gaps are not abstract. A general-purpose AI receptionist is typically built to answer questions from a knowledge base and route everything else. That works reasonably well for store hours and directions. It falls apart the moment a caller wants to do something, which on a service line is most of the time.

Booking a service appointment requires knowing what slots are actually open in the DMS right now, which advisor or bay capacity applies, whether the job is a quick lube or a diagnostic that needs a longer block, whether a loaner or shuttle is needed, and whether the vehicle has an open recall that should be added to the visit. A caller who says she needs her “A service” or that her “maintenance minder is at 15 percent” is speaking a language that dealership staff understand instantly and that a generic model will often misread. A system without live DMS access cannot confirm a time; it can only collect a message or point the customer to the website she has already tried. At that point the AI has not reduced the advisor’s workload. It has added a step in front of it and handed the customer a reason to call somewhere else.

There is also a quieter reason generic deployments disappoint in dealerships. Experienced advisors and BDC agents compensate for a lot of missing structure: they know which tech handles diesel work, which customers always want the first slot, and which recall parts are actually in stock. That knowledge rarely lives in a system. An AI that is dropped in without being built around how a service department actually operates has nothing to compensate with, so it fails in exactly the places the team was quietly covering.

What Is at Stake When the Phone Fails

The phone remains the primary channel for fixed ops. A Marchex Institute analysis of more than 8 million calls to U.S. dealerships found that nearly 70 percent of inbound calls were about parts and service, and that more than 19 percent of all calls went unanswered or were abandoned. More than 10 percent of the callers who abandoned did so during automated routing or voicemail. In other words, dealerships already lose customers to phone systems that make callers work too hard. A generic AI agent that cannot book, cannot see the schedule, and cannot transfer with context does not fix that problem; it simply adds a more conversational version of it. Each of those callers represents a repair order, a customer-pay or warranty opportunity, and a relationship that is far easier to keep than to win back.

What Purpose-Built Voice AI Looks Like in Practice

When AI on the service line is designed around resolution instead of deflection, the experience changes for both the customer and the team. The customer from the opening scenario calls at 8:40, describes her check engine light, and the agent recognizes this as a diagnostic visit rather than routine maintenance. It checks live availability, offers two times that fit, confirms her vehicle and contact details, notes the open recall on her VIN, and sends a confirmation text before the call ends. If she asks something that genuinely needs an advisor, such as a question about a prior repair or a goodwill request, the call is warm-transferred with a summary of who she is and what she needs, so the advisor picks up where the conversation left off instead of starting over.

The measure of success also shifts. A dealership should judge AI on the phone by booked appointments, first-call resolution, and how rarely a customer has to repeat themselves after a transfer, not by containment rate. A system that keeps callers away from advisors while those callers quietly book elsewhere can look efficient on a dashboard and still be costing the service lane revenue every day.

How Viova Approaches the Service Line Differently

Viova was built specifically for automotive dealerships, which is the difference that most of the criticism in the broader AI conversation points toward. Viova answers inbound calls when the team is busy or the store is closed, understands fixed ops language, and connects directly to the dealership’s DMS so it can book, reschedule, and cancel appointments against live availability. When a call needs a person, Viova routes or warm-transfers it with the context already captured, and it sends confirmation texts so customers leave the call with something concrete in hand.

Viova does not replace staff. It removes the pressure from high-volume call handling so the team can focus on high-value conversations. During the morning drop-off rush, that means advisors stay with the customers standing in front of them while the phone is still being answered and appointments are still being booked. After hours, it means the service lane keeps filling its schedule for the next day instead of collecting voicemails that someone has to return at 7:30 the following morning.

For any dealership evaluating voice AI, the useful questions are the ones generic tools struggle to answer well. Can it read and write to the DMS in real time, or does it only take messages? Does it understand how the service department actually describes jobs, packages, and visit types? What happens when a caller asks for a person, and what does that person see when they pick up? Who reviews real calls after launch, and how quickly are gaps addressed? A vendor that can answer those questions specifically is offering something very different from a chatbot with a voice.

The Problem Was Never the Voice

The research and the commentary point to the same conclusion. Customers are not rejecting AI on the phone because it is AI; they are rejecting systems that make them repeat themselves, cannot complete the task they called about, and stand between them and a person who can. On a dealership service line, avoiding that outcome comes down to whether the AI understands the business, has access to the systems that run it, and knows when to bring in the team. Missed calls are new revenue opportunities, but only if the system answering them can actually turn a call into a booked appointment.

Book a Demo to hear how Viova handles a real service call, from first ring to a confirmed appointment in your DMS.