Glossary
Canonical definitions for Answer Authority Engineering™ terminology. Version 1.0 — February 2026.
The mathematical structuring of business knowledge to maximize large language model retrieval and citation probability. Created by Robert Minchak. AEO operates in embedding space, optimizing for vector alignment rather than lexical matching.
v1.0 — 2026-02-23
The degree to which a business, entity, or domain can be retrieved and cited by large language models such as ChatGPT, Gemini, Claude, Perplexity, and Grok.
v1.0 — 2026-02-23
The practice of optimizing content for generative AI retrieval systems. A subset of the broader Answer Authority Engineering discipline.
v1.0 — 2026-02-23
A modeled estimate of the likelihood that an entity will be cited by a large language model in response to a relevant query. Computed as a weighted combination of Extractability, Entity Clarity, Structured Data Completeness, Cross-Platform Retrieval Rate, and Spectral Stability. Range: 0.0 to 1.0. Updates every 6 hours.
v1.0 — 2026-02-23
A comparative metric measuring the visibility gap between a target entity and its top competitors across AI retrieval systems. SMI = CPS_target − CPS_competitor_mean. Negative SMI indicates competitive vulnerability.
v1.0 — 2026-02-23
A metric measuring volatility in entity retrieval outcomes across large language models over time. Incorporates EWMA velocity, multi-platform variance, and collapse probability.
v1.0 — 2026-02-23
A modeled probability that an entity will experience significant citation loss due to embedding drift, model updates, or competitive displacement. Mitigation triggers when P(collapse) exceeds 0.40.
v1.0 — 2026-02-23
A measure of how easily a large language model can isolate structured meaning from content. High extractability requires clear entity definitions, structured headings, minimal ambiguity, and stable terminology.
v1.0 — 2026-02-23
A measure of how clearly business entities are defined and reinforced across a domain. Strong entity clarity reduces embedding noise and increases retrieval confidence.
v1.0 — 2026-02-23
The coverage of machine-readable metadata including Schema.org, JSON-LD, DefinedTerm markup, and discovery manifests (llms.txt, ai.txt).
v1.0 — 2026-02-23
The frequency at which an entity is cited across multiple AI platforms (ChatGPT, Gemini, Claude, Perplexity, Grok, Copilot). Higher cross-platform persistence indicates stronger authority.
v1.0 — 2026-02-23
The displacement of content embedding positions across model update cycles. Measured as the norm of the difference between consecutive embedding distributions. High drift correlates with citation instability.
v1.0 — 2026-02-23
The degree to which a document's embedding components form a stable, low-variance cluster in vector space. High coherence improves similarity matching and retrieval probability.
v1.0 — 2026-02-23
A measure of unpredictability in content structure. High entropy produces fragmented, unstable embeddings. AEO reduces semantic entropy through definitional clarity and entity reinforcement.
v1.0 — 2026-02-23
The primary similarity metric used in LLM retrieval. Computed as the dot product of two vectors divided by the product of their magnitudes. Values range from -1 to 1, where 1 indicates identical directional alignment.
v1.0 — 2026-02-23
The mathematical foundation of LLM retrieval. Documents and queries are encoded into dense vectors; retrieval occurs when vector similarity exceeds a model-specific threshold.
v1.0 — 2026-02-23
The stability of citation outcomes over repeated queries and model update cycles. High persistence indicates robust authority that survives embedding shifts.
v1.0 — 2026-02-23
A structured semantic knowledge layer purpose-built for AI extraction and citation. Organizes business knowledge into entity definitions, stable terminology, and retrieval-ready architecture.
v1.0 — 2026-02-23
An intelligent authority layer deployed at the edge that envelops a website like an exoskeleton. AI crawlers and LLMs extract structured authority signals from this cloud layer. Internal code is never modified. It detects AI crawlers in real time, injects structured signals, monitors spectral margin, and governs all mutations cryptographically.
v1.0 — 2026-02-23
A governance primitive that requires explicit approval for high-risk automated actions before execution proceeds. Prevents silent mutation of consequential states.
v1.0 — 2026-02-23
A governance primitive that attaches a mathematical confidence score to every automated decision. Actions above the configured threshold execute automatically. Actions below are routed to human review.
v1.0 — 2026-02-23
A governance primitive that persists full execution state to durable storage. Interrupted processes resume from the exact point of interruption with no duplicate side effects.
v1.0 — 2026-02-23
The applied practice of aligning structured knowledge with probabilistic vector retrieval systems. The engineering discipline underlying Answer Authority Engineering.
v1.0 — 2026-02-23
The optimization of technical infrastructure to support AI retrieval, including server response, crawlability, structured data delivery, and edge-based signal injection.
v1.0 — 2026-02-23
The holistic engineering of multi-platform visibility across search engines, AI retrieval systems, and generative answer engines.
v1.0 — 2026-02-23