TL;DR: Fox Motors' CIO tested multiple AI vendors across 44 dealerships. The five questions that mattered most: What happens after the call is answered? Does it empower your team or replace them? How does it enforce follow-up? Can it handle messy data? And does it truly integrate with your DMS? The answers will tell you more than any demo.
Every vendor at every conference has a demo that looks impressive. The voice sounds natural. The appointments get booked. The dashboard is beautiful.
And then you turn it on.
Yuri Demidko has been through this cycle. As the CIO of Fox Motors, he's responsible for technology across 44 franchise dealerships. He built custom software to audit their phone traffic and discovered what most dealers already suspect: over 80% of calls at some locations went straight to voicemail, and more than half of those voicemails were never returned.
So he went looking for AI that could actually solve it. He tested multiple providers. Some of them worked. Partially. What he found is that the difference between a good demo and a good result comes down to a handful of questions most dealers never think to ask.
Here are five of them.
Every AI vendor can answer a phone call. That's table stakes in 2026.
The harder question is what happens sixty seconds later. Does the information from that call go somewhere useful? Does anyone act on it? Is there a system that ensures the customer actually gets helped, or does the data just land in an inbox nobody checks?
Yuri put it this way: "We actually tried a number of different companies. They handled the front end of it really well. The thing is, on the back end, that's where the real difference was."
The "front end" is the AI picking up the phone and having a conversation with the customer. Most vendors nail this. The "back end" is what happens next: Does the information get routed to the right person? Does it create a task? Does it connect to an open repair order?
If you're evaluating a vendor and the demo ends when the call ends, you're only seeing half the product. Ask what happens in the five minutes after the conversation. Across 1,000+ dealerships, we've seen that the back end is where revenue gets rescued or lost.
This is the question that separates dealerships that adopt AI successfully from the ones that fight it for six months and cancel.
Fox Motors doesn't have AI answer every call. They give their service advisors three rings. If a human picks up, the customer gets the personalized experience the dealership is known for. If nobody can get to it, because the advisor is with a customer on the drive or the phones are stacked on a Monday morning, the AI picks up on the third ring.
"We're not trying to replace the people with the AI," Yuri said. "What we are trying to do is use AI as a stopgap for things that people could not get to."
This matters because it changes how your team receives the technology. When AI is positioned as a safety net instead of a replacement, adoption happens faster. Nobody feels threatened. The advisor juggling a customer at the counter and three ringing phone lines sees AI as the thing that stopped them from losing a call they couldn't physically answer.
Ask your vendor: Can we control when the AI engages? Can we set it up as a fallback instead of a front line? And what does the advisor experience actually look like? Does it feel like help, or like oversight?
This was the insight that surprised Yuri most. The AI's biggest value wasn't answering calls. It was making sure the humans did their part afterward.
Here's how it works at Fox Motors: When a call comes in and information is gathered (whether by the AI or through a text conversation), it gets assigned to the right advisor and the right repair order. If the advisor doesn't follow up within 30 minutes, the system nudges them. If they still don't follow up, it escalates to the service manager, the general manager, whoever the dealership has designated.
"That is where the true impact was," Yuri said. "I remember that whole stat where we had an issue with people not following up. Literally, that went away. I've had unsolicited feedback from multiple general managers telling me: what's crazy is I have not had customers call me and complain that they can't reach anybody or get the information they need."
Most AI vendors can pick up the phone. Far fewer can tell you whether your team followed through, and do something about it when they don't.
When you're evaluating, ask: What's the accountability loop? What happens when my advisor doesn't respond? Who gets notified, and when?
Yuri's background is in technology. He came from Apple before moving into automotive, and he made a point on the podcast that most dealers miss entirely.
He told a story about business intelligence tools. A vendor comes in, shows you a beautiful dashboard. The graphs are moving, you can drill into everything, it looks nothing like the reports you're pulling from your DMS. You sign up. They turn it on. And every number is wrong.
"The vendor didn't lie to you," Yuri said. "You just took their tool and hooked it up to your data, and your data has duplicates. Every customer apparently owns 47 vehicles. People who passed away are still getting flyers."
The same problem applies to AI. "You take this super intelligent AI," he said, "and you hook it up to dirty data, and it will provide the most robust, eloquent, beautifully worded wrong answer of all time."
He called it AI slop. The quality of any AI tool is only as good as the data underneath it. If your CRM hasn't been cleaned in years, if your DMS has duplicate records stacked on duplicates, the smartest AI in the world will just give you confident-sounding nonsense.
Before you sign with an AI vendor, ask: How do you handle data quality? Do you audit our data before going live, or do you just plug in and hope? What's your approach when the data doesn't match reality?
The vendors worth working with will talk about data first, AI second.
"Integration" is the most overused word in vendor pitches. Every company claims they integrate with your DMS. But there's a difference between reading data from your system and actually working inside it.
Yuri described what real integration looks like at Fox Motors: When a customer calls about an open repair order, the AI doesn't just log the call. It identifies the repair order, assigns the follow-up to the correct advisor, and creates a task tied to that specific job. The advisor doesn't have to open a separate app, cross-reference a call log, or manually update anything. The work shows up where they already work.
Integration that saves time looks different from integration that adds another screen to check.
Here's why it matters at scale: An advisor with 20 customers in front of them can communicate with 100 customers via text through a tool that's embedded in their workflow. But if that tool requires them to toggle between systems, you lose the efficiency gain. Yuri made this point directly: "An advisor with 20 people lined up can communicate with 100 people via text. But they can only have a conversation with one of those 20 people or that one phone call."
The tool has to multiply their capacity, not fragment their attention.
When evaluating, ask for a walkthrough of the advisor's day, not the admin dashboard. Where does information land? How many clicks to act on it? Does it live in the DMS, or does it create another tab?
Every one of these comes back to a single principle that Yuri articulated early in the conversation: "It's not necessarily that people are bad. It's more just there's so many calls coming into the service department that the advisors are obviously dealing with people in front of them. Things go to voicemail. Things get forgotten."
Your people are doing the best they can inside a system that was never designed for the volume they handle today.
The right AI vendor understands this. They're building a system that makes it structurally difficult for things to fall through the cracks: calls get answered, tasks get assigned, follow-ups get tracked, and accountability lives inside the workflow instead of getting bolted on after the fact.
If a vendor can't answer these five questions clearly, that's useful information. And if they can, push further. Ask to talk to a CIO or service director at one of their existing dealerships. The best proof is always in the follow-through.
*Hear the full conversation between Yuriy Demidko (CIO, Fox Motors) and Numa CEO Tasso Roumeliotis on the *Millionaire Car Salesman Podcast.
No more hold music. No more unanswered voicemails. Your customers are top priority.
