Your Lead Gets Four Different Names Before Noon. The AI Never Mentions It.
The 10 p.m. Tahoe Problem
Here's a scenario every GM has lived, whether they knew it or not: a shopper messages your website at 9:57 p.m. asking about a Tahoe. An AI assistant books her for 10 a.m. the next day. She shows up. The salesperson has no idea who she is, what she asked, or that the appointment even exists.
Durran Cage, founder of Cage Automotive, put a number on this mess in a recent CBT Live segment: his mystery shopping found an average of four different "names" representing the dealership in a customer's first 72 hours — the AI, an automated text, a BDC rep, a salesperson, sometimes a manager confirming on top of all of it. Nobody owns the thread. The customer just sees a dealership that can't keep its story straight.
This is the part the vendor demos never show you, and it's worth dragging into the light before you sign another AI contract.
The fix isn't more AI — it's a playbook. Cage's prescription is almost boringly operational: define who does what, when, built around the AI vendor's actual process, your CRM, and your human team's real workflow. The AI should identify itself as a virtual assistant and always offer a path to a person. A manager should be reading AI conversation logs regularly, not just the close rate. And handoff time — not just response time — needs to be tracked and reported. Cage's benchmark: dealers should be converting 30–40%+ of leads to appointments, and they should know precisely how many of those the AI booked solo versus with human help, broken out by lead source. Most stores can't currently answer that question. That's the tell.
The Evidence Is Mixed, Which Is More Honest Than It Sounds
The industry's own numbers back up why this matters right now. One recent survey found 87% of dealers already use AI in some form, but only 6% have a fully defined set of rules governing it — 24% have no oversight at all. That's not a technology problem. That's a management vacuum with a chatbot standing in it.
And the stakes of getting it wrong are concrete, not theoretical. A dealer chatbot in Toronto offered a customer $7,000 over market on a 2021 BMW X3. The customer didn't know they were talking to AI. When the store tried to walk it back at the in-person meeting, the deal nearly died — and only media attention forced the dealer to honor the original number. That's the downside case for letting an AI negotiate unsupervised with no human spot-checking the output.
On the other side of the ledger, there's real signal that a well-managed handoff works. monday.com's new Harmony voice platform points to one actual customer, a UK car finance broker, reporting a 9% increase in closed deals after AI started following up on approved applications the moment they came in — qualifying about half the conversations and handing 80% of qualified ones live to a human adviser "with a full briefing." That last phrase matters more than the 9%: the AI isn't just passing a name and number, it's passing context. That's the difference between a handoff and a hand-wave.
Similarly, Sanas just launched "Supervised AI," built explicitly around the idea that a human should govern up to four AI-run calls simultaneously from one console — an implicit admission from a voice AI vendor that fully unsupervised calling isn't where the trust level is yet, and shouldn't be.
What This Means for Your Store
The AI-answers-everything pitch is no longer the interesting part — every vendor does that now, from inride's trade-in bot to HubSpot's sales agents. The differentiator is whether a dealership can prove its handoff doesn't drop the customer on the floor. Before adding another AI tool, audit what your current ones already do: how many "names" does a lead encounter before a human actually engages? Does anyone review the transcripts? Is handoff time a tracked metric, or just response time?
The stores that win this next cycle won't be the ones with the flashiest bot. They'll be the ones where the bot and the floor act like they work for the same business — because right now, for a lot of dealers, they don't.
Flai Raises $27M. The Phone Still Breaks If Nobody's Listening.
Flai closed a $27M Series A after its AI agents started booking 50,000 dealership appointments a month and posting a 20x revenue jump, with backers including Toyota Ventures and two dealer groups putting real money behind it. But the same week, a veteran consultant warned that unsupervised AI phone agents are quietly costing stores service business when calls go sideways and nobody's checking the recordings.
Flai Answers the Phone 50,000 Times a Month. Your Service Advisor Still Has to Believe It.
Flai just raised $27 million after proving it can book 50,000 dealership appointments a month without a human picking up the phone. The money is real, the appointment count is real — but so is the dealership down the road where nobody noticed the AI mispronouncing the store's own name for months. Here's what separates a vendor pitch from a tool your front counter actually trusts.