In Part 1, we looked at why an AI-generated rental rate can appear credible and still produce an incomplete or commercially inaccurate comparison. Now comes the harder question: How should the industry determine whether an AI-presented offer can actually be trusted?

The Standard: AI Rate Integrity

The car rental industry needs a way to discuss the quality of AI-presented offers. I use the term AI Rate Integrity:

AI Rate Integrity is the degree to which an AI-assisted shopping experience accurately represents the price, availability, terms, source, comparability, and book-ability of the offer an operator intends to sell.

I think of this as two related ideas. AI Rate Integrity is the standard. AI Distribution Intelligence is how we measure and monitor it over time, across markets and platforms. That means watching how AI systems discover, interpret, compare, rank, and present an operator and its offers. Commercial search coverage is one of the tests inside that standard.

AI Rate Integrity can be evaluated through six tests:

·         Accuracy: Does the displayed rate match the cited or linked source?

·         Freshness: When was the offer last verified, and how likely is it to have changed?

·         Completeness: Does the price include mandatory taxes and fees? Are important conditions and exclusions clear?

·         Comparability: Are the dates, location, vehicle class, length of rent, inclusions, and booking terms truly equivalent?

·         Coverage: Did the AI inspect a meaningful portion of the relevant market before making a ranking or recommendation?

·         Book-ability: Can the traveler still complete the transaction at the represented price?

A result can pass one test and fail another. A rate may be accurate but stale. It may be current but exclude mandatory fees. It may be fully bookable but drawn from such a narrow source set that the word "cheapest" is not justified.

AI visibility and AI rate integrity are not the same discipline.

Who Owns the Answer?

Inside a rental company, the ownership is not obvious. Revenue Management owns pricing strategy and competitive position. Distribution owns channel connectivity and rate delivery. Marketing owns content and discoverability. E-commerce owns conversion. Technology owns systems and access.

Outside the company, OTAs and brokers are responsible for the accuracy of the offers they expose, and AI platforms determine how sources are selected, combined, and presented to the user.

Each party owns part of the chain. No one clearly owns the complete journey from the operator's intended rate to the traveler's AI-generated answer.

That may be the biggest risk. Problems that sit between departments often remain invisible until they affect bookings, trust, or revenue.

The answer will be cross-functional, but cross-functional cannot mean ownerless. I believe Revenue Management is the natural convener. It already owns competitive position. It already runs the rate shop. And it is the function that feels the consequence first when a traveler is steered to another supplier. Revenue cannot fix every link in the chain, but it can own the question, the measurement, and the escalation when the answer is wrong.

The 10-Minute AI Rate Audit

Operators do not need to wait for a new industry standard to begin learning. A basic audit can be run today, in about the time it takes to review a rate shop:

1. Select one important airport or city market.

2. Choose two pickup dates.

3. Choose two lengths of rent.

4. Choose two vehicle classes.

5. Run the same prompts across several AI systems.

6. Record which suppliers appear, how they are ranked, which rates are shown, and which sources are cited.

7. Check taxes, fees, class comparability, booking terms, and whether the offer remains available.

8. Compare the result with the live rate the operator intended to sell.

9. Repeat the same audit weekly.

The purpose is not to catch one assistant making one mistake. The purpose is to identify patterns.

Which sources appear most often? Is the direct site ever considered? Do the same competitors consistently appear? Are certain classes mapped incorrectly? Does the operator appear for "best value" but disappear for "cheapest"? Are quoted rates current and bookable?

I suspect most operators who run this exercise once will discover something uncomfortable: they know exactly what rate they sent into distribution, and almost nothing about what an AI-assisted shopper actually saw.

That gap is AI Distribution Intelligence in its simplest form. Operators already apply this discipline to rate shops and channel performance. The AI layer is simply the newest place it is needed.

There is more to say about the mechanics than one article can hold. In a forthcoming Auto Rental News article, Julie Flores will go deeper into how these agent questions are actually answered, how to measure effectively how your brand appears in the results, what tools are available today for that measurement, and most importantly, what operators can do to effect change. Consider the audit above the starting point that work builds on.

What Happened When We Checked the Real Market?

We have started running these checks ourselves. One of the more interesting results came from Atlanta. The AI agent got two important parts of the request right: the market was ATL, and the vehicle class matched the car type we asked for. At first glance, the answer looked credible.

But the reference rates were for the wrong month.

At first, that may sound like a small error. To a revenue manager, it is not. The rental dates define the demand period, availability, competitive set, and price being compared. A rate can be perfectly accurate for the date it represents and still be completely wrong as an answer to the traveler's question.

This is the kind of failure that may be harder to spot than an obviously bad answer. The location was right. The car type was right. The rates looked plausible. A traveler could reasonably assume the comparison matched the dates they requested. It did not.

That is why AI Rate Integrity has to evaluate the whole commercial request. An answer can look right in several dimensions and still be commercially wrong. Market, dates, length of rent, vehicle class, taxes, terms, source, freshness, and book-ability all have to line up closely enough to support the claim being made.

The ATL test also keeps this in perspective. AI travel shopping is advancing quickly, and many answers are already useful. There is still some distance between a useful conversational answer and a commercial comparison that a revenue manager would treat as a reliable rate shop.

Hotels Just Showed the Industry the Timeline

Agentic booking is no longer a forecast. In August 2026, Expedia announced that travelers in the United States can book Expedia and Hotels.com hotel inventory directly inside AI Mode in Google Search. A traveler asks a conversational question, compares options with real-time pricing, guest reviews, and cancellation policies, selects a room, and completes the booking through Google Pay with immediate confirmation. Expedia and Hotels.com own the reservation and the customer service from there.

Until this change, AI Mode could compare options but had to hand the traveler to a third party to book. Now the transaction is completed inside the answer.

There are real advantages to this approach. The pricing is live and supplied by the booking partner rather than scraped from the open web, so freshness and book-ability, two of the six tests, are addressed by the design. The transaction also has a named owner. The traveler knows who confirmed the booking and who to call when plans change.

But there is a tradeoff. The comparison set is defined by commercial integration, not necessarily by how widely the agent searched. Two brands, both part of one company, supply the inventory, yet a traveler may reasonably read the result as a market comparison. Suppliers outside the integration, including direct channels, may never enter the consideration set.

And the ranking logic inside the answer is not visible to the suppliers being ranked. Visibility in this model is no longer earned only through content and optimization. It can depend on who holds a distribution deal with the answer layer.

For car rental, the important point is what this tells us about timing. Hotels are in some ways a more complicated product than car rental, with thousands of differentiated properties, room types, and amenity packages, and hotel choice is certainly more subjective.

The best hotel is partly a matter of taste. If conversation-to-confirmation booking is already live for a product that complex and that subjective, it is hard to argue that car rental, a more standardized product that travelers shop largely on price, vehicle class, and terms, is far behind.

If hotels are already this far along, car rental should assume it is next.

The Next Competitive Set May Not Be on the Rate Shop

Travel companies are building AI into planning, partner tools, traveler experiences, and now the transaction itself. Expedia Group has also emphasized that trusted data, reliable inventory, and accurate pricing remain central to AI adoption. Both signals point in the same direction.

Today, a revenue manager asks:

·         Where do we rank on the OTA?

·         Who moved above us?

·         Which competitor lowered its rate?

·         Should we respond?

Tomorrow, the same revenue manager may also need to ask:

·         Which suppliers did the AI consider?

·         Which channels did it inspect?

·         Why did it rank one offer ahead of another?

·         What rate did it show?

·         Was our company omitted?

·         Which offer did the agent recommend?

And eventually:

·         Which offer did the agent choose to book?

At that point, AI has moved beyond a marketing or search issue. It is part of distribution and revenue management.

At RateHighway, we have spent more than two decades helping operators decide what rate to send into the market. Increasingly, I find myself asking what rate comes back out.

Operators don't need to wait for a formal industry standard to get started. Revenue Management can begin now:

1. Measure what AI systems show travelers.

2. Track source coverage as well as rate accuracy.

3. Verify dates, class comparability, total price, terms, and book-ability.

4. Assign an internal owner for AI Rate Integrity and a process for escalating discrepancies.

For decades, our industry has focused on one fundamental question:

·         What rate should we send to the market?

AI introduces a second:

·         What rate did the market tell the traveler we have?

Agentic commerce introduces a third:

·         What rate did the traveler's agent choose?

AI is already becoming part of distribution. The car rental industry should decide who owns the integrity of that chain before ownership is assigned by default.