This is the first article in a two-part series on how AI interprets and displays car rental rates differently than traditional channels, and how operators should adapt.

I recently asked an AI assistant a very simple travel-shopping question: “Who has the cheapest midsize rental car at LAX next Friday?”

The answer was careful. It named two travel sources, discussed estimated prices, and suggested checking live booking sites before making a reservation.

Here is what stayed with me. LAX is one of the largest and most competitive rental markets in the country. A traveler shopping it may encounter direct brand websites, franchise websites, major online travel agencies, broker sites, metasearch platforms, and affiliate channels. Well over a dozen relevant sources compete for that one booking.

The assistant surfaced two.

That left me with a more important question than whether the quoted price looked reasonable:

How much of the relevant market did the AI actually inspect before answering?

The question matters because the answer to any "cheapest" prompt depends entirely on the scope of the search behind it.

"Cheapest among the sources I checked" is not necessarily the same as "cheapest."

If an AI assistant inspects only a small portion of the market, it may still return a reasonable answer. What it may not have is enough evidence to make a definitive market claim.

This is not a criticism of one AI platform or one travel source. It is a structural question for the car rental industry.

When AI begins comparing, ranking, recommending, and eventually booking dynamically priced rental products, who is responsible for the quality of the commercial answer?

The Distribution Chain Just Got Longer

For decades, car rental operators have worked to improve a distribution path that is complicated but familiar:

Revenue strategy -> Rate engine -> Reservation system -> OTA, broker, or direct channel -> Traveler

Each step carries its own risks. Rate codes must map correctly. Vehicle classes must align. Taxes and fees must be handled properly. Availability must be current. Channel rules and booking conditions must survive the journey.

AI adds another layer:

Revenue strategy -> Rate engine -> Reservation system -> OTA, broker, or direct channel -> AI interpretation layer -> Traveler

That new layer does not simply display a rate table. Depending on the system and task, it may rewrite the traveler's question, choose which sources to inspect, summarize what it finds, compare offers, and present a recommendation. OpenAI's own documentation, for example, describes ChatGPT Search rewriting user prompts into targeted web queries and drawing on search partners, and describes its agent mode navigating websites and taking actions on a user's behalf.

And this is already moving beyond interpretation. In August 2026, Expedia announced that U.S. travelers can book Expedia and Hotels.com inventory directly inside AI Mode in Google Search. The answer layer is beginning to become a transaction layer. Car rental should treat that as a timetable, not a distant scenario.

That changes the question from "Did we send the correct rate?" to "What did the AI understand, and what did it tell the customer?"

Visibility is the First Question, not the Last

Generative Engine Optimization, or GEO, is a useful part of this discussion. GEO asks whether an AI system can find a company, understand its content, and represent it in generated answers. A recent car rental industry white paper took a useful look at this issue, including how often brands appear in AI answers, which sources are cited, and how closely AI-quoted rates match direct rates.

For car rental, that work matters. A supplier that does not appear in the answer may never enter the traveler's consideration set.

But visibility is only the first question. The next questions are more operational:

·         Did the AI find the current rate?

·         Did it compare equivalent vehicle classes?

·         Did it include mandatory taxes and fees?

·         Did it understand the booking conditions?

·         Did it inspect enough relevant sources to support its ranking?

·         Can the traveler still book the offer it presented?

A brand can be visible and still be represented incorrectly. An individual rate can be accurate and still support an incomplete market conclusion.

Cheapest of What?

This is where the scope of inspection becomes commercially important.

Suppose an AI assistant checks three sources and finds Supplier A at the lowest price. The answer may be correct within those three sources. But if the market contains many other relevant channels, what exactly should the assistant say?

"This is the cheapest option" is a broad commercial claim.

"This is the lowest price I found among the sources I checked" is a narrower and more transparent claim.

The distinction may sound small, but it can move real revenue from one company to another.

In ordinary informational search, incomplete source coverage may not change the answer. In commerce, source coverage can determine the winner.

I believe the industry will need a better understanding of commercial search coverage, meaning the breadth and relevance of the sources an AI system evaluates before making a comparative claim. Coverage is not a separate discipline. It is one part of the broader standard I call AI Rate Integrity, which I will define more fully below.

Travelers may eventually need clearer answers to questions such as:

·         How many sources were checked?

·         What types of sources were included?

·         When were the prices verified?

·         Were direct sites included, or only third-party aggregators?

·         Were any relevant sources unavailable to the agent?

Without that context, the user sees the conclusion but not the strength of the evidence behind it. The rental operator may not even know whether its offer was considered.

In Defense of a Simple Question

Some readers will bristle at building this discussion around "who is the cheapest." Our industry has a complicated relationship with that question, especially when automation is involved. When rates fall, software is often blamed for moving too fast or chasing competitors downward.

I have heard that argument for more than two decades. The problem is that automation does not set strategy. It executes one. Rates move because of competitive posture, demand, utilization, and fleet position, including the condition Richard Lowden described in his closing keynote at this year's International Car Rental Show in Dallas: a market "carrying more fleet than it can sell." The software makes the response faster and more visible, but the underlying cause is still a business decision or a market condition.

So why use "cheapest" as the test case? First, because travelers will ask it. Revenue management may be sophisticated, but the public's prompt will often be one sentence. Operators do not get to choose the questions travelers type.

Second, "cheapest" is measurable. Define the market, dates, length of rent, vehicle class, taxes and fees, and timestamp, and the claim has a correct answer that can be checked and repeated. "Best" is subjective. "Cheapest" can be audited.

You cannot audit "best." You can audit "cheapest."

That makes it a useful test for AI shopping. When an AI system declares a winner, the productive response is to ask what it actually measured: which sources, which dates, how fresh the rates were, whether the offers were comparable, and whether the result was still bookable.

The danger is not the simple question. It is a confident answer nobody can measure.

Car Rental Makes the Comparison Unusually Difficult

A rental car is not a static item with one universal price. Before two offers are truly comparable, several distinct groups of variables have to line up:

·         The rental itself: pickup and return location, pickup and return time, length of rent, and vehicle class.

·         The money: taxes, airport fees, deposit requirements, and payment terms.

·         The conditions: mileage allowance, fuel policy, cancellation rules, and included or excluded products.

·         The clock: demand, utilization, and availability, all of which move constantly.

Even the phrase "midsize car" can require interpretation. The traveler speaks in natural language. The reservation and distribution systems may rely on specific class definitions and SIPP codes. Two offers that look similar in a conversational answer may not be commercially equivalent.

Then there is time. A rate may be accurate when the agent collects it and unavailable minutes later because inventory moved, demand changed, a promotion ended, or the vehicle class closed.

That means a responsible commercial answer needs more than a number. It needs context.

Should AI Agents Back off From Certain Claims?

I do not think the answer is that AI should stop helping travelers shop. These systems can make a complicated process much easier.

But there is a meaningful difference between discovery and declaration.

Consider the progression:

1. "These rental companies serve LAX."

2. "I found these current offers."

3. "These offers differ in price."

4. "This is the cheapest offer."

5. "This is the best choice for you."

6. "I booked this offer for you."

Each step carries more commercial influence than the one before it.

My view is simple:

The stronger the commercial claim, the stronger the verification requirement should be. An agent does not always need to refuse an answer. It may need to qualify it.

A more responsible response could say:

These are the best current offers I found from the sources available to me. I checked three sources. Prices and availability can change, and the final total should be verified before booking.

That language is less absolute, but more useful. It tells the traveler what the system knows and what it does not know.

There is also an ethical question. When an AI answer can influence where a person spends money, should the system disclose uncertainty more clearly than it would for a general informational answer?

I believe it should.

The larger risk is false confidence in a comparison that may be incomplete.

Part 2 is coming soon.

Author's bio: Michael Meyer is president of RateHighway, a revenue optimization technology company built specifically for the car rental industry. He has spent more than two decades working with car rental operators on pricing, market intelligence, distribution, and revenue automation. For more than 23 years, RateHighway has helped operators understand the market and control the rate they send into it. The next challenge, he believes, is understanding the rate the market sends back to the traveler.