AI Operations · Glossary
Vector Retrieval
Vector Retrieval inside AI Operations is one input source to Dynamic Context Engineering, not a standalone capability. It retrieves over embedded representations under authored governance for residency, freshness, and index scope.
Vector Retrieval embeds documents, policies, and knowledge artefacts into a vector index and retrieves them by similarity to a query embedding. Inside an Interactive Agent, Vector Retrieval is one of several input sources Dynamic Context Engineering composes from. It is not a substitute for the Glossary component, for Policies, or for authored Variables. It supplies background knowledge where authored vocabulary and authored rules are insufficient. The retrieval pattern is governed: which indices, which embeddings, which residency, which freshness. The retrieved items appear in the production trace alongside the rest of the context that produced the decision. Vector Retrieval defined as a standalone product is not Production AI.