A Standard Should Say When It Changes
brand.context v2.0 is live. Here is why it was revised, and what's actually changed in the market since April
We published brand.context in April, a machine-readable standard for brands to declare purchase-decision evidence to AI systems. We're publishing version 2.0 today. This note explains why, plainly, rather than letting the change go unremarked.
Why It Changed
Version 1.0 was built on an earlier internal taxonomy, eight evidence filters labelled T1 through T8. That language has since been retired in our own published research in favour of the Linkage Gap and Reasoning Gap framework, a sharper, more rigorously tested distinction between evidence that changes a model's recommendation and model judgments that no evidence will override. Publishing new research under one framework while an older, contradictory standard stayed live under the old one was a real inconsistency, and worth fixing in the open rather than quietly.
A standard that revises itself without saying so is not a standard worth trusting. Version 2.0 states the change directly, in its own first section, and Version 1.0 stays archived at its original DOI, marked clearly as historical, not current.
What Changed Around Us
The revision also landed at a genuinely different moment than April. In the past few weeks, Google shipped two new specifications for machine-readable infrastructure, the Open Knowledge Format and Agentic Resource Discovery. An independent open standard called EntityMap, backed by well-known names in entity SEO, reached stable version 1.0 and ran a public consultation. Ahrefs published data showing 97% of LLMs.txt files, an earlier and simpler machine-readable format, received zero AI requests after publication.
The web is growing a second layer, written for machines. Almost none of it is being read yet.
That last point matters more than the new specs themselves. Publishing a file is not the same as being consumed, for any standard, including this one. Version 2.0 says so explicitly, citing the Ahrefs figure and the early, unconsumed state of comparable standards, because a standard that only admits its limits when asked is not being straight with the people adopting it.
The Commercial Context, Stated Plainly
brand.context is published by AIVO, a company that sells diagnostic and remediation services built on the evidence this schema defines. That is worth saying outright rather than leaving implicit. It is also, as it turns out, the normal state of this entire emerging category. EntityMap is authored by the founders of Waikay, a commercial AI brand-monitoring product in broadly the same space as our own, and Waikay is named as EntityMap's own reference implementation. Every open standard currently shipping in this space right now sits next to a commercial product built by the same people who wrote it.
That reframes the question worth asking about any of these standards. It is not which one has no commercial interest attached, since none currently do. It is which one is grounded in the most rigorously tested evidence about what actually changes an AI system's recommendation. That is the basis on which we'd ask brand.context to be judged, not a claim to neutrality we don't have and aren't making.
What's Actually New in v2.0
Alongside the terminology correction, three changes: every evidence field now carries a confidence marker, separating independently verifiable claims from a brand's own self-declared ones. Every field is tagged with whether it addresses a closable retrieval gap or a structural model judgment evidence won't override, an honesty check against overselling what this schema can do. And the standard now states formal conformance rules, with a validator and public comment process explicitly named as commitments in progress, not existing infrastructure dressed up as finished.
What This Means for Brands Evaluating Any of This
Adopt what's cheap to adopt. Don't expect any single file, ours or anyone else's, to be a guaranteed channel into how these models recommend a brand today. Ask what evidence a standard's claims actually rest on before adopting it as strategy. brand.context v2.0 is our answer to that question for ourselves, published openly, including the parts that are still unfinished.
AIVO Standard