Technical guide · original methodology
What does machine-native mean for a cannabis website?
A technical definition based on explicit information, bounded interfaces, and evidence—not a promise of rankings or automated sales.
A useful definition
In this observatory, machine-native describes a website whose public information and supported capabilities can be interpreted without guessing at hidden state. Cannabis is our first research cohort because it exposes difficult questions about age boundaries, public information, and trustworthy interpretation. The definition is architectural. It does not certify legality, product safety, eligibility for advertising, or suitability for a particular person.
Start with information, not protocols
A small information site may need accurate HTML, stable links, clear organizational identity, and consistent structured facts. It may have no reason to operate an agent endpoint. Adding MCP to an ordinary website does not repair a confusing identity, an inaccessible article, or a stale policy page. We propose documenting each public representation and comparing its claims against the same authoritative record.
Keep five claims separate
Discoverable means a resource can be found. Reachable means an identified observer retrieved it under stated conditions. Understandable means a representation contains the facts necessary for a defined task. Authorized means a caller has permission for a specific action. Effective means the work produced a measured outcome. These are separate hypotheses. A sitemap supports discovery; it cannot establish any of the other four on its own.
What an operator can test
Choose an informational page and record its canonical URL, retrieval timestamp, response type, visible heading, declared entity, and public machine representations. Ask whether a second person can repeat the observation. Where a gate prevents inspection, record the limitation. Where ownership logs are needed, leave the result unknown. The deliverable is a reproducible evidence record, not a badge that says the whole business is AI-ready.
What success would mean
A useful first milestone is a smaller set of contradictory public facts and a clearer boundary around what remains unobserved. Subsequent authorized measurements can examine actual requests, referrals, and account activations. Google documents no special AI file or schema requirement for its search AI features. The commercial case for this work must therefore rest on reliable operations and measured outcomes, not a supposed secret ranking signal.
Sources and scope
Google: AI features and your website · AIWebSignals: current public capability contract
Provider documents support the referenced technical facts. The interpretation, examples and proposed checks are AIWebSignals methodology, not provider endorsement or measured industry results.
Review the assessment method and its limits · Request a correction privately · Read the JSON version