AODM (AI-Optimized Data Markup) is a lightweight, extensible schema that makes websites and datasets inherently understandable to AI systems — from search engines and agents to large language models and generative tools.
Just as SEO transformed how content is discovered on the web, AODM structures data at the source so that machines do not have to guess at it. For website owners and developers that means better visibility and stronger performance in AI-driven search and discovery. For anyone building on that data, it means less preprocessing and fewer silent errors.
The problem it solves
Knowledge today moves between systems as prose, ad-hoc JSON, or vectors in a database, and almost all of the context is lost in transit. A number arrives without its unit. A statement arrives without its source. A fact that expired two years ago arrives looking exactly like one confirmed this morning. Each consuming system reinvents its own conventions for patching this up, and none of them interoperate.
HTML succeeded because it solved a universal problem — how to represent and link documents — before any one browser dominated. AODM is aimed at the equivalent problem for AI-era knowledge: how do facts, entities, and their relationships move cleanly between databases, language models, graph engines, retrieval pipelines, engineering systems, and agents, without becoming specific to any one vendor or model?
The design
The core is deliberately small: four knowledge primitives —
entity, relationship, fact and
rule — and three annotations — source,
confidence and value. That is the whole vocabulary. A
system that implements those seven elements can read any conformant AODM document.
Everything that makes the data trustworthy is carried as optional attributes on those same elements rather than as new tags. This matters more than it sounds: a parser that does not understand validity dates or derivation still reads the document correctly, so the format can carry rich data without punishing simple consumers.
Design rule: the number of elements stays small; the expressiveness of those elements does not have to. Adding an element enlarges the surface every implementer must handle. Adding an optional attribute does not.
Two ways to use it
Standalone documents. An <aodm:knowledge> file, in
XML or JSON, describing a body of knowledge in its own right — suitable for
exchange between systems, for archiving, or for feeding a retrieval pipeline.
Embedded in HTML. The v1.1 embedding profile annotates existing web pages in place, so the content a visitor reads and the data a machine reads are the same content and cannot drift apart. It remains supported alongside 1.2.
Open by design
The specification, both schemas, the validation rules, and the conformance tests are published together. Nothing about AODM requires this project’s code: the documents are the definition, and anyone can build a compatible implementation from them. That is a deliberate choice — a format that only one toolchain can read is not a standard.
Contact us at support@nesaraq.com.