AI voice agents at car dealerships: inbound, outbound, and what they must disclose

The question worth asking about a voice agent on a sales line is not whether it sounds human. It is which calls it is allowed to touch. A dealership that answers that question narrowly gets a useful tool. A dealership that answers it broadly gets a system talking to buyers about money with nobody watching, which is a different product with a different risk profile.

Inbound: only the calls nobody answered

Every store has a set of calls that go nowhere. Sunday at two. Tuesday at 6:40 when the two people on the floor are both with customers. The third transfer of a call that started in service. The line that rings out because the receptionist stepped away. Those calls are not currently being handled well by a human, because they are not being handled by anyone.

That is the right place for a voice agent, and it is a narrower place than most deployments choose. The rule is simple: if a person on your team was available, the person takes the call. The agent exists for the overflow and the after hours window, and the store should be able to prove that from the routing configuration rather than from a promise.

Scoped that way, the agent has a short job. Get the caller's name and number early, because everything else can be recovered if you have those two. Answer the question that was actually asked, if it is a factual one: hours, whether a specific unit is still available, where the store is, what is needed for an appraisal. Offer a specific time. Write what happened into the record so the person who follows up starts with context.

What the inbound agent should not do

• Take a call a person could have taken. Overflow is a routing decision, and it should be configured, not assumed.
• Negotiate. Price, payment, trade allowance, term, fees. Any figure that moves is a person's job.
• Commit. Holding a unit, promising delivery, confirming a rate, guaranteeing an appraisal figure.
• Keep going when the caller is confused, upset or asking for a person. All three are handoff triggers.
• Collect credit or payoff information. Auto dealers that arrange financing sit inside federal rules on safeguarding customer information. Decide deliberately what any automated channel may touch.

Outbound: draft, then approve

Outbound is where voice agents go wrong in retail, because volume is easy and judgment is not. The design that holds is draft and approve. The system assembles the call list and the reason for each call. A manager reads the list, removes the names that should not be called, and releases the rest.

This is not a performance tax. It is the only version where a store can answer the question a customer will eventually ask, which is why did your system call me. If the answer is because a rule fired, you have a problem. If the answer is because a manager reviewed it Tuesday morning and approved it, you have a process.

Approval also catches the cases that rules never catch cleanly. The customer who bought last week from a different store and told you so. The one in a dispute with service. The one who asked to be left alone in a text three days ago that the calling system never saw. A manager scanning a list catches those in seconds.

What the law says about an artificial voice on the phone

This is general information rather than legal advice, and the details are fact specific and state specific. Have your own counsel review any voice deployment before it dials.

The Telephone Consumer Protection Act, at 47 U.S.C. 227, restricts calls made with an automatic telephone dialing system and calls delivering a message using an artificial or prerecorded voice, and the FCC's implementing rules appear at 47 CFR 64.1200. In a declaratory ruling adopted on February 2, 2024, in CG Docket No. 23-362, the FCC confirmed that the restrictions on artificial or prerecorded voice cover current AI technologies that generate human voices, which means such calls fall under the TCPA and require the prior express consent of the called party absent an emergency purpose or an exemption.

The rules also carry identification requirements. Under 47 CFR 64.1200(b), an artificial or prerecorded voice message must state clearly, at the beginning, the identity of the business responsible for initiating the call, and must state a telephone number for that business during or after the message. Telephone solicitations to a residential subscriber may not be initiated before 8 a.m. or after 9 p.m. local time where the customer is. The FTC's Telemarketing Sales Rule sets the same outbound calling window at 16 CFR 310.4 and carries do-not-call obligations of its own.

Three consequences for a dealership. An inbound caller who dialed you is a different consent situation from an outbound call you placed to them, so treat the two separately from the first day. Whatever consent your lead forms and text programs actually capture is a question for counsel, not for a vendor's marketing page. And state law can go further than federal law, on recording, on disclosure and on automated calling.

Disclosure, in practice

Beyond what the rules require, disclosure is a product decision that determines whether customers trust the store afterward. Three behaviors hold up.

1. Answer the question honestly, immediately. When a caller asks whether they are speaking to a person, the agent says it is an automated assistant, in plain words, without a deflection and without a joke.
2. Do not simulate a named employee. An agent that introduces itself as a person on your staff creates a problem the day the customer walks in and asks for that person.
3. Say what happens next. A caller who understands that a salesperson will call back in the morning behaves very differently from one who thinks they just handled it.

Start by counting what you are actually missing

Most stores buy a voice agent before they know the shape of the problem it is supposed to solve. Spend a week finding out first, because the answer changes what you buy and sometimes removes the need to buy anything.

Pull your inbound call log and count unanswered calls by hour and by day of week. Separate three categories that look identical in a report and are completely different in practice: calls that rang out with nobody on the floor, calls that rang out while three people were standing at the desk, and calls that were answered but abandoned on hold or lost in a transfer. The second and third categories are staffing and process problems, and no agent fixes them. The first is the window a voice agent is for.

You may find that your missed call volume is concentrated in eleven hours a week. That is a much smaller and much safer deployment than answering everything, and it is the one to start with.

Permissions are the real specification

Whatever an agent is technically allowed to do, it will do, at three in the morning, on the call you did not anticipate. So write the permissions as a list and treat that list as the product.

• Which lines and which hours it answers, and the routing rule that puts a person first.
• What it may state as fact: hours, location, availability, process.
• What it may never state: price, payment, trade value, rate, availability guarantees.
• What it may write into your systems, and what it must leave for a person.
• Which conditions force an immediate handoff or a callback commitment.
• What happens when it fails, because it will: a person, a voicemail to a monitored box, and a call in the queue the next morning.

What to do before you turn anything on

1. Count your missed calls by hour and day, so you know the size of the actual problem.
2. Configure routing so a person is always offered the call first.
3. Write the permission list and the handoff triggers before you write the script.
4. Ask counsel about consent, disclosure and recording for your states.
5. Listen to every agent call for the first two weeks, without exception.