Delivery-layer fixation designates, in Pagup’s vocabulary, the risk that an observed AI answer is not a fresh model generation, but a reconstruction delivered by an application: cache, routing, prompt template, approved answer, orchestration or older retrieval state.
Why the term matters
An organization may correct its site, sources and proof while still observing an older answer in a third-party application. The problem may not come from the current corpus. It may come from the surface delivering the answer.
This distinction prevents two objects from being confused:
- the native reconstruction, obtained when a model or controlled layer is interrogated directly;
- the delivered reconstruction, obtained inside a final application that may have its own caches, thresholds, prompts, sources and freshness policies.
How Pagup uses it
Pagup uses it as a methodological guardrail in AI perception audits. When an answer repeats unusually or appears decoupled from current sources, the analysis must avoid concluding too quickly that the answer is faithfully stable.
The correct diagnostic first qualifies the observed object: native model, delivered surface, enterprise assistant, chatbot, answer engine, retrieval-enabled interface, or application with unknown delivery layer.
For the primary doctrinal definition, see gautierdorval.com. For normative qualification, see interpretive-governance.org.