Google Just Published How It Wants AI to Read a Business. Your Dealership Website Is Still Written for Humans.

On June 12, Google Cloud quietly shipped something that will never make the automotive trade press, has nothing to do with cars, and doesn't mention websites once. It's called the Open Knowledge Format, or OKF, and on the surface it's a plumbing project for data engineers.

Ignore it at your own risk. Because buried in a post about internal company data is the clearest picture yet of how Google thinks AI should read a business — and it looks nothing like the website you're paying for today.

Here's what Google actually announced. OKF is, in their words, "a vendor-neutral, agent- and human-friendly standard for representing the metadata, context, and curated knowledge that modern AI systems need." In plain English: it's a simple way to hand an AI a clean, organized map of everything an organization knows. Technically it's almost boringly simple — a folder of plain-text files, one per concept, each tagged with a type and linked to the others so the whole thing forms a connected graph. A "customer" file links to an "orders" file links to a "product" file. The AI doesn't read a document top to bottom and guess. It walks the map.

That's the tell. Google — the company that decides whether shoppers ever find your dealership — just published its preferred blueprint for machine-readable knowledge. And the blueprint is built on entities and the relationships between them, not pages of prose.

We've Seen Google Do This Before

If this feels familiar, it should. This is the third act of a play Google has run twice already, and both prior runs minted winners and losers in search.

Over a decade ago, Google, Microsoft, Yahoo, and Yandex — bitter competitors who agreed on almost nothing — jointly launched Schema.org, a shared vocabulary for labeling the things on a web page so search engines could understand them: this is a price, this is a review, this is a vehicle, this is a business address. At the time it sounded like optional busywork for developers. It wasn't. The businesses that adopted structured data early started winning rich results, star ratings, and knowledge panels while everyone else stayed a blue link. The ones who waited spent years catching up.

The pattern is always the same. Google signals what "structured" means. A window opens where early adopters get outsized visibility. Then structure becomes table stakes, and being unreadable becomes a penalty rather than a missed bonus.

OKF is that signal for the AI era. Google isn't asking you to publish OKF files — the format is aimed squarely at internal company data, the schemas and metrics and runbooks locked in the heads of senior engineers, and it says nothing about SEO or public websites. But it tells you, in Google's own hand, what a machine wants when it's trying to understand something: not a beautifully written page, but a clean set of connected facts.

Your dealership is a set of connected facts. The question is whether an AI can see them.

What This Means for Your Dealership

Right now, a shopper doesn't type "Honda dealer near me" and scroll. They ask ChatGPT, Gemini, or Google's AI, "Which dealership near me has a low-mileage certified CR-V under $30k and good service reviews?" — and the AI answers before your homepage ever loads. To be in that answer, the AI has to be able to read your dealership the way OKF reads a company: as entities it can connect.

Here's the uncomfortable part. Most dealer websites are still built like a brochure — designed for a human to look at, not for a machine to parse. The critical facts about your store are there, but they're trapped:

• Your inventory is a photo carousel and a price that loads in JavaScript a crawler may never execute, instead of clean, linked vehicle records with year, trim, mileage, drivetrain, and options as data.
• Your service department is a paragraph of marketing copy, when it should be structured facts: what you service, hours, what a given repair costs, how to book.
• Your financing, trade-in process, staff, and location live in prose scattered across pages that don't reference each other — so nothing connects. The AI can't tell that this service special applies to that vehicle you sell at this address.

An AI trying to recommend you is doing exactly what OKF describes: looking for typed entities and the relationships between them. When your site is a pile of unconnected pages, the machine has to guess — and machines that guess about your dealership either get it wrong or skip you for a competitor whose facts are legible. That's not a traffic problem you'll see in a rankings report. It's a "you were never in the conversation" problem, and it's invisible until you go looking for it.

The dealers who show up in AI answers aren't the ones with the prettiest sites. They're the ones whose inventory, services, and reputation are structured cleanly enough that a machine can read them without guessing.

A Word of Caution Before the Pitches Land

Within a few months, a vendor is going to email you promising to "make your dealership OKF-ready" or "publish your OKF knowledge graph for AI." When that pitch lands, delete it.

OKF is version 0.1, explicitly "a starting point, not a finished standard," and it was built for internal data pipelines, not public websites. No AI shopping assistant is reading dealer OKF files today, and nobody can promise you it will. Paying to bolt a brand-new internal-data format onto your website is the 2026 version of every "secret Google format" scam that came before it.

The signal is real. The specific format is not your homework. Don't confuse the two.

What To Actually Do About It

You don't chase the acronym. You do the durable work the acronym is pointing at — making your dealership legible to a machine. Concretely:

• Get your inventory into structured data, not just pixels. Every vehicle should carry its real attributes — year, make, model, trim, mileage, price, VIN, drivetrain, key options — as machine-readable fields with proper vehicle schema, not just an image and a caption. This is the single highest-leverage thing most dealers are missing.
• Turn your service, financing, and trade-in pages into clear facts. Replace vague marketing copy with plainly stated, structured answers to the exact questions shoppers ask AI. If a human has to interpret it, a machine will too.
• Connect the entities. Link your vehicles to the financing and service that apply to them, your specials to the models they cover, everything to your real location and hours. Relationships are half of what OKF is about, and they're what lets an AI answer a specific question with your dealership.
• Fix the fundamentals that block reading in the first place. Fast pages, clean markup, content that renders without a shopper — or a crawler — waiting on heavy scripts. The best structured data in the world is worthless if the AI can't load it.
• Audit what the machines can actually see. You cannot fix what you can't measure. Find out today which of your facts an AI can read and which are invisible to it — because that gap is the exact shape of the AI answers you're being left out of.

The Bottom Line

Google didn't announce a website standard on June 12. It did something more useful for anyone paying attention: it wrote down, in the open, how it believes AI should read a business. Entities. Types. Relationships. A clean map, not a pretty page.

Every prior time Google told the web what "structured" means, the businesses that listened early won a window of visibility, and the ones who waited paid to catch up. The AI-search era is that window, open right now. Your competitors' websites are just as unreadable as yours today. The advantage goes to whoever gets legible first.

If you want a straight answer on what an AI can and can't see when it reads your dealership right now, book a no-pressure walkthrough and we'll show you exactly where you stand — and what it would take to become the answer instead of a page nobody's machine can read.