The AI Pilot Phase Is Ending at Dealerships. The Phone Is Where Outcomes Get Measured.

The AI headlines aimed at customer service have gotten specific. Gartner projects that by 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention, cutting operational costs by roughly 30%. It also expects 40% of enterprise applications to ship with task-specific AI agents by the end of 2026, up from less than 5% in 2025. Read those two forecasts together and the direction is hard to miss: the software your store already runs on is about to start answering, routing, and resolving on its own.

For a dealership, though, forecasts do not set appointments. The useful question is not whether agentic AI is coming to customer service - it plainly is. It is narrower and more practical: where should a dealer let it prove itself first, on a workflow where the result is visible in 60 days and easy to walk back if it underperforms?

The answer most operators land on is the phone. Here is why, and how to judge it like an analyst rather than a buyer of hype.

Myth: AI phone agents are mainly about cutting your BDC headcount. Data: Most dealerships cannot hold the headcount they already have. Total dealership turnover reached 42% in 2024 - a three-year high - and sales-department roles churned at roughly 60%, per the NADA Dealership Workforce Study. AI's first job on the phone is not replacement. It is continuity.

The hype is finally being graded

For two years, "AI at the dealership" mostly meant experimentation: a chatbot here, an email generator there. That phase is ending. In Cox Automotive's AI Readiness in Auto Retail study (October 2025, based on 537 franchise dealership leaders), 81% of dealers said AI is here to stay and 63% said investing now is critical to long-term success. But only about 15% had actually embedded AI into workflows and decision-making. Belief is nearly universal; operational adoption is not.

The gap is the story. Lori Wittman, president of retail solutions at Cox Automotive, put it plainly: "Dealers don't care about AI for AI's sake. They care about outcomes they can measure - more cars sold, lower inventory costs, higher gross profit." The same study found the friction behind the caution: 74% of dealers worry about AI accuracy and errors, and 66% want more education before they lean in. In short, 2026 is the year dealers stop grading AI on ambition and start grading it on results.

That reframes the buying decision. The question is no longer "Is this impressive?" It is "Can you show me the number this moves, and can I watch it move?"

Why the phone is the first real test

Most dealership AI use cases are hard to score. Did an AI-written email actually sell a car? Did a predictive model change what you stocked? The causal line is fuzzy, which is exactly why so many pilots stall.

The inbound phone is the opposite. It is one of the most measurable operations in the building:

• Answer rate: calls answered versus missed, by hour and by department.
• Speed to answer: how long a high-intent caller waits before a human, or nobody, picks up.
• After-hours capture: what share of calls arrive when the store is closed, and what happens to them.
• Appointments set: the outcome that ties a call to the showroom floor or the service lane.

Every one of those is a hard number you can baseline this week and re-check in 60 days. And the stakes are concrete: the cost of a missed call at a dealership is not a missed click - it is frequently a ready-to-buy or ready-to-service customer who simply dials the next store. Notably, when Cox asked dealers where they were putting AI to work first, the top use case was not flashy content generation. It was engaging customers 24/7 with real-time automated text, chat, or email (52%) - the always-on layer. The phone is that same instinct applied to the highest-intent channel a dealership has.

The trajectory dealers are buying into

Source: Gartner press releases, August 2025 and March 2025.

None of that means a dealership should hand its phones to a fully autonomous agent tomorrow. It means the tooling, the vendor maturity, and customer expectations are all moving the same way - and the dealers who learn to measure and manage it now will be fluent when it becomes table stakes.

The staffing math nobody writes on the board

Return to turnover, because it is the quiet reason the phone problem never fully resolves with hiring. When a dealership's sales and BDC roles turn over at 40-60% a year - and 2024 saw sales-consultant turnover jump about 13 percentage points, one of the sharpest increases in years (NADA) - the phone is being answered by a team that is perpetually half-new. New reps ramp slowly, miss more calls, and qualify less consistently. Every vacancy is a stretch of dropped or fumbled calls that never shows up as a line item on a financial statement.

An always-on AI layer changes that math not by replacing the team but by stabilizing it. It answers when everyone is with a customer, after close, and on the Saturday you are short-staffed. It captures the name, the vehicle of interest, and the callback time; it books the appointment, or routes the hot one to a human immediately. The measurable result - fewer missed high-intent calls, more appointments set - is exactly the kind of outcome the Cox study says dealers now demand before they will believe.

How to adopt it like an analyst

The same Cox research offers a warning worth heeding: AI "can't be just another point solution," and the average dealership already juggles more than 40 software systems. Bolting on one more disconnected tool is how pilots die. A measured path looks like this:

1. Pick one workflow where the score is visible. Inbound sales or service calls, especially after-hours and overflow, is the classic starting point.
2. Baseline before you buy. Pull your current answer rate, after-hours call volume, and appointment-set rate. If you cannot measure it now, you will not be able to prove the lift later.
3. Insist on guardrails. Because 74% of dealers worry about accuracy, demand clean escalation to a human, transparency that callers are speaking with an assistant, and logs you can audit.
4. Give it 60 to 90 days and read the numbers. Expand only where the data earns it.

That is the un-hyped version of AI adoption - and, not coincidentally, the version dealers say they actually want.

The bottom line

The forecasts about agentic AI in customer service are real, and they point one direction. But a dealership does not need to bet the store to participate. It needs one workflow where outcomes are legible, reversible, and tied to revenue. The phone fits that description better than any other: high intent, easy to measure, and chronically undermanned. Prove it there, and the rest of the roadmap gets a lot easier to read.

If you are mapping where an always-on phone layer fits your store, that is the problem Carbuki works on - AI voice agents built for dealership sales and service calls. You can learn more at carbuki.com.

Sources

• Gartner, "Gartner Predicts Agentic AI Will Autonomously Resolve 80% of Common Customer Service Issues Without Human Intervention by 2029" (March 5, 2025): gartner.com
• Gartner, "Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026, Up from Less Than 5% in 2025" (August 26, 2025): gartner.com
• Cox Automotive, "Automotive Dealers Are Ready for AI to Deliver Outcomes and Skip the Hype, According to New Cox Automotive Study" (October 28, 2025): coxautoinc.com
• NADA, "NADA Dealership Workforce Study" (2025): nada.org