Google announced the Open Knowledge Format (OKF), a new specification for representing organizational knowledge so it can be shared with AI agents, tools, and organizations. OKF uses markdown files, making it easy for both humans and machines to read, analyze, and use.
What Problem Does OKF Solve?
AI systems often need knowledge that exists outside the model, but that knowledge is often scattered across catalogs, wikis, repositories, shared drives, and other internal systems. Google Cloud’s announcement says OKF is meant to address that problem by representing scattered knowledge in a common format that can be read and used by humans, AI agents, tools, and organizations.
The announcement explains the reason for developing this new specification:
“As foundation models continue to improve, the lack of relevant context often limits what they can do, especially as they are used to build agentic systems. While these models can help you write code, summarize documents, or analyze a dataset, they still need the right information to produce accurate and actionable results. “
What Open Knowledge Format Is
OKF is a format for representing organizational knowledge in a way that can be shared with AI agents, tools, and organizations. The format represents datasets, metrics, APIs, and tables as markdown files that can be read by both humans and AI systems.
A non-jargon way to think of it is that there are OKF documents, which are the trees, and OKF bundles, which are the forests. An OKF document can be a markdown file with a name like orders.md or weekly_cart_abandonments.md.
OKF is designed to be a simple format for representing knowledge that can be used across different platforms, allowing the same knowledge to be shared between AI agents, LLMs, tools, and organizations.
The announcement explains:
“To make the format concrete, we’re publishing reference implementations at both the producer and consumer ends:
- An enrichment agent that walks a BigQuery dataset, drafts an OKF concept document for every table and view, then runs a second LLM pass that crawls authoritative documentation and enriches each concept with citations, schemas, and join paths.
- A static HTML visualizer that turns any OKF bundle into an interactive graph view in a single self-contained file; no backend, no install on the viewing side, no data leaves the page.
- Three ready-to-browse sample bundles: GA4 e-commerce, Stack Overflow, and Bitcoin public datasets, produced by the reference agent and committed to the repo as living examples of conformant OKF.
These are proofs of concept, deliberately. The agent demonstrates one way to produce OKF; nothing about the format requires a specific agent framework or LLM. The visualizer demonstrates one way to consume it; nothing about the format requires HTML or a graph view. We expect (and want!) the ecosystem of producers and consumers to grow far beyond what we’ve shipped.”
Who OKF Is For
The OKF standard is written for two groups of users: those who create OKF bundles (producers) and those that consume them (consumers).
- Producers:
The people and systems that create and maintain the knowledge, such as documentation and internal policies. - Consumers:
The systems that use that knowledge, such as AI agents, LLMs, a search index, or a visualizer.
The above list gives examples of OKF bundle producers and consumers.
Who Consumes OKF and Why?
AI agents and LLMs are among the primary consumers of OKF. They use the format to access the information and curated knowledge needed to perform tasks and produce accurate results.
Contexts In Which AI And LLMs Can Use OKF
- Coding agents
- Data analysis agents
- Research agents
- Internal enterprise assistants
- Agentic workflows
How Does OKF Structure Its Files?
OKF uses markdown files and YAML frontmatter, making the format readable and editable by people using standard tools. YAML frontmatter is metadata placed at the very top of the file. It is written as key-value pairs separated by a colon, such as author: Anne Example. This makes OKF easy for humans to read and understand, making it useful to people who create, edit, maintain, or work with organizational knowledge.
People Who May Create, Edit, Maintain, Or Work With Organizational Knowledge
- Software engineers
- Data engineers
- Analytics teams
- Technical writers
- Business teams
How Can Organizations Use OKF?
Organizations can use OKF to represent and share institutional knowledge that would otherwise remain scattered across shared drives, tools, and other repositories.
Organizations That May Find OKF Useful
- Organizations building AI agents
- Data teams
- Engineering teams
- Knowledge management teams
Availability
Google is proposing a common format for representing organizational knowledge rather than a new platform for managing it. The OKF specification, reference implementations, and sample bundles are available on GitHub.
The announcement says that it is a starting point:
“OKF v0.1 is a starting point, not a finished standard. The format will evolve as more producers and consumers emerge and as we collectively learn what knowledge representations agents actually need in practice.
We’re publishing in the open from day one because that’s the only way a knowledge format earns its name, whether you’re building a knowledge catalog, an enrichment pipeline, a wiki tailored to AI agents, or anything in the AI knowledge domain.”
An explainer tweet by Tech with Mak shared why OKF solves a problem:
“The most underrated idea in agent tooling this year might be a gist Andrej Karpathy wrote about “LLM Wikis” – markdown libraries that agents read, update, and maintain on their own.
What followed was predictable. Teams everywhere started building their own version – AGENTS[.]md, CLAUDE[.]md, Obsidian vaults wired into coding agents, folders of index[.]md and log[.]md files agents consult before doing anything.
…Google just tried to close that gap with the Open Knowledge Format – a spec that says => here’s the one field every concept needs (type), here’s a small set of optional fields if you want them queryable, and otherwise, write however you want.
It’s not a new tool or platform. It’s an agreement on shape, which is exactly what Karpathy’s pattern needed to stop being a hundred incompatible reinventions of the same idea.”
Read the original announcement here:
Introducing the Open Knowledge Format
Featured Image by Shutterstock/Poetra.RH
