Break the question into observable requirements
A question such as whether a company can serve a market contains dimensions: which capability is needed, what geographic scope matters, for which period and what evidence could support the conclusion. Making these requirements explicit separates available data from missing conditions. A contract prevents an ambiguous question from silently becoming a simplified choice that the current state cannot justify.
Sufficiency changes by dimension
The same evidence may be relevant to confirm that an organization published a certificate and insufficient to claim that the certificate covers a particular facility today. The answer depends on the predicate, subject and temporal scope. Knowledge should not be labelled sufficient or insufficient as one block; each dimension keeps its own state, requirements and explanation.
Check before asking a model to make a judgment
An evaluator can compare options when the context contains the information those options require. If a critical fact is missing, a fluent answer does not resolve the gap. The AnswerabilityGate preserves why the state is insufficient and prevents model probability or confidence from substituting for evidence. When a contract is answerable, the result remains a replaceable structured evaluation, not canonical truth.
Answer what is known and bound the rest
Answerability does not force a choice between a complete answer and total silence. It may allow a partial result: answer the supported part, mark another as UNKNOWN and identify evidence needed to continue. This makes the response useful without extending it to a broader fact. It also allows reassessment when information, contract or period changes.
Conceptual example
Hypothetical: a source confirms that an organization published a certificate but does not say which site it covers or whether it remains current. The observation of the publication can be answered. A question about the present certification of a specific plant lacks sufficient answerability until suitable scope and time are available; a model should not infer them from the document title.
Scope and limits
Answerability is not a business quality score, total certainty or guarantee that sources are correct. It depends on an explicit contract, available state and exact question. An answerable question may still contain uncertainty, contradictions or source limits. If requirements are unmet, UNKNOWN or abstention is valid and does not mean FALSE.
Editorial basis
This page explains product doctrine and boundaries. It does not demonstrate source coverage or observed outcomes.
Read the product modelExplore the Knowledge library
Explore AXIGNAL