Analyze consistency and determinism in outputs across runs, temperatures, and variations.
Measures how much output varies across repeated runs of identical input. Establishes whether a system is dependable enough for a given use, and is frequently skipped.
Anywhere consistency is itself a requirement — regulated processes, and any context where two users asking the same question must receive the same answer.
Creative applications where variation is the point.
For processes subject to audit, expected variance must be documented. An auditor asking why the same query produced different answers on two dates needs an answer prepared in advance.
Testing variance at low temperature during evaluation and running at higher temperature in production, so the measured consistency describes a configuration nobody uses.
Running an identical reconciliation explanation request twenty times and finding that the cited evidence is stable while the narrative wording is not — which is the acceptable shape of variance here.
The source material did not expand this solution into components, and the row says so in its own notes. Expanding it here would be authorship, not research.
Two different absences share this shape. Foundational solutions get built whatever the domain, so no domain links them; the rest are solutions this domain genuinely does not reach for. v_ai_solutions_unlinked separates the two.