Data infrastructure and fiber-optic cables illustrating knowledge sharing between AI agents via Google Cloud's Open Knowledge Format

In short, On June 12, 2026, Google Cloud published the Open Knowledge Format (OKF), an open, vendor-neutral standard for describing an organization's knowledge, datasets, metrics, APIs, runbooks, as markdown files readable by a human and by any AI agent. The near-term stake is not public search rankings, but a deeper trend: knowledge that isn't structured and named becomes invisible to AI. The same reflex that made schema.org win on the web now extends to business knowledge.

On June 12, 2026, Sam McVeety (Tech Lead, Data Analytics) and Amir Hormati (Tech Lead, BigQuery) announced on the Google Cloud blog the release of the Open Knowledge Format, version 0.1. The specification, its reference implementations and concrete samples were available from day one on the GoogleCloudPlatform/knowledge-catalog GitHub repository. Google explicitly describes it as "a starting point, not a finished standard."

The problem OKF tackles is concrete. "The lack of relevant context often limits what [foundation models] can do, especially as they are used to build agentic systems," Google writes. In plain terms: a brilliant AI agent is useless if it knows nothing about your product catalog, your internal metrics or your business rules. OKF proposes a standard wrapper for that knowledge.

What the format contains

Technically, OKF is deliberately minimal: a directory of markdown files with YAML frontmatter, versionable in Git, readable on GitHub, consumable by any agent. No proprietary database, no mandated platform. Google insists it is "a format, not a platform," independent of the cloud provider, the AI model or the agent framework (ADK, LangChain or homemade).

June 12date Google Cloud published OKF v0.1
v0.1framed as a "starting point," not a frozen standard
3sample bundles shipped: GA4 e-commerce, Stack Overflow, Bitcoin
2reference implementations: BigQuery agent + HTML visualizer

To seed adoption, Google ships three ready-to-browse sample bundles (public GA4 e-commerce, Stack Overflow and Bitcoin datasets) and two reference implementations: a BigQuery enrichment agent and a static HTML visualizer. In parallel, Google's Cloud Knowledge Catalog is updated to ingest OKF bundles directly.

Why an SEO publisher should care

At first glance, OKF is about enterprise data, not search. But the underlying signal is exactly the one we track in GEO: AI favors structured, dated, named information. When Google standardizes how agents consume knowledge, it validates a way of reading that applies to your public content too.

It is the direct continuation of what we've documented for months. Zero-click search already reaches 68% of queries, and each AI engine builds its own citation logic, as shown by the fragmentation of market share between ChatGPT, Gemini and Claude. OKF adds a piece to the puzzle: knowledge becomes a raw material that machines exchange, and only well-packaged knowledge travels.

The Cicéro angle, OKF is not a new ranking lever to flip on tomorrow morning. It is a direction signal. Google is telling us plainly that the future belongs to machine-readable knowledge. Schema.org structured data did this for your pages; OKF does it for internal knowledge. The lesson is the same: what isn't structured doesn't get cited.

What it changes for your visibility

You don't need to publish your pages in OKF to rank well: the format targets companies' internal agents first. But the logic it enshrines applies to any content meant to be picked up by an AI. Three concrete actions follow.

  1. Break key content into self-contained, citable blocks. A direct answer, a definition, a named figure: that's what an agent extracts. A paragraph buried in filler doesn't surface.
  2. Date and source everything. OKF formalizes metadata (who, what, when). Apply the same rigor to your pages: visible date, named source, attributed data. That's what makes content trustworthy to an AI.
  3. Keep your entity consistent. Same name, same identifiers, same data across your site, your AI performance reports in Search Console and your documents. That consistency is what lets a machine connect the pieces and cite you with confidence.

Is your knowledge readable by AI?

Cicéro audits your GEO visibility and produces the structured content that gets you cited by Google AI Mode, ChatGPT and Claude. Agency-quality work, software-grade productivity, from €250 to €1,800 per month.

What this article does not cover

OKF is at version 0.1: its real reach will depend on adoption by players other than Google, which remains to be proven. This article does not detail how to write an OKF bundle or integrate it with an agent framework, the full spec lives in the GitHub repository. Crucially, Google makes no explicit link between OKF and search ranking: the visibility angle developed here is a Cicéro reading of the trend, not a Google promise. Finally, no usage or adoption data is available yet, the format having just been published.

Our take

Google isn't trying to re-rank the web with OKF, it is laying a deeper marker: making structured knowledge the raw material of agents. For a business leader, the message is simple. The work that makes your knowledge machine-readable is no longer a technical detail; it decides whether an AI cites you or ignores you. OKF won't change your SEO tomorrow; it confirms the direction to invest in today.

Frequently asked questions

What is Google's Open Knowledge Format (OKF)?
An open, vendor-neutral specification published by Google Cloud on June 12, 2026. It describes an organization's knowledge (datasets, metrics, APIs, runbooks) as markdown files with YAML frontmatter, stored in a Git-versionable directory. Goal: the same knowledge base can be read by a human and consumed by any AI agent, without proprietary translation.
Does OKF replace schema.org for SEO?
No. Schema.org describes a web page's content for search engines; OKF describes an organization's internal knowledge for AI agents. The two are complementary. The underlying signal is shared: structured, dated, named information is more easily picked up by an AI than vague text.
What should you actually do about OKF?
Nothing urgent to publish in OKF for organic visibility: the format targets internal agents first. But the direction is clear. Structure your key content into self-contained, citable blocks, date it, name your sources and figures, and keep entity consistency across your site, your schema.org and your documents.

Sources

Alexis Dollé, founder of Cicéro
Alexis Dollé
CEO & Founder

Growth and SEO content strategist, I founded Cicéro to help businesses build lasting organic visibility — on Google and in AI-generated answers alike. Every piece of content we produce is designed to convert, not just to exist.

LinkedIn