Google Releases Open Knowledge Format for AI Website Interpretation

Image credit: Search Engine Journal

Google published its Open Knowledge Format (OKF) on June 13, 2026, offering a potential solution for public websites to create structured knowledge graphs that could enhance how AI agents interpret online content.

The format, initially designed for internal company knowledge, allows concepts within linked Markdown files and YAML frontmatter to form a network of relationships, providing AI agents with a deeper understanding beyond traditional flat page consumption.

Google described OKF as being composed of “just markdown,” “just files,” and “just YAML frontmatter,” according to Sani, highlighting its simplicity and cost-effectiveness due to the absence of a runtime, SDK, or build step.

Although developed for internal use (v0.1), Sani successfully applied OKF to the “No Hacks” website, demonstrating its viability for public web content.

Implementing OKF for public websites necessitates maintaining a separate copy of content, which presents a synchronization challenge, but also encourages clearer articulation of a website’s underlying knowledge, Sani reported.

This approach could evolve the existing ‘llms.txt’ identity file into a comprehensive map of a website’s knowledge, allowing AI agents to query information directly rather than relying on scraping, according to Sani.

The format aims to overcome limitations of current web content consumption by AI, which often struggles with the context and relationships between disparate pages.

Source: Search Engine Journal