
Technical leaders · Scientific & technical
Making organizational knowledge usable by AI agents
A practical look at the Open Knowledge Format and why portable context, metadata and documentation matter for more dependable AI agents.
A language model can be powerful and still misunderstand an organization’s terms, data ownership and decision rules when the right context is missing. The Open Knowledge Format (OKF) proposes storing that context as simple Markdown files with YAML metadata, making knowledge readable by people and portable across agent systems.
A practical lesson for data teams
The useful idea is separation from a particular tool. Metric definitions, table descriptions, usage limits and update dates can travel with the data instead of being trapped in one vendor’s catalog. A sensible pilot starts with one bounded domain—sales or supply chain, for example—then records its glossary, authoritative sources and the owner responsible for reviewing each document.
OKF is an early specification, so it should be treated as an experimental documentation contract rather than a finished standard. Its value still depends on content quality, version control and access controls. The format does not correct poor knowledge; it makes well-managed knowledge easier to exchange and consume.
Source: Google Cloud — Introducing the Open Knowledge Format