The AI Reputation vocabulary.

What AI assistants say about a person or business when asked. The new substrate of trust in 2026, partially replacing Google search results and online reviews. Unlike traditional reputation, AI reputation is private — the buyer reads the answer alone, with no review thread to respond to and no SERP to monitor.

A 0–10 composite measurement of how well AI engines recognize and accurately describe a person or business. MirrorAI scores across four dimensions — recognition, accuracy, completeness, and citation quality — across all five major engines. Full definition on the methodology page.

The discipline of shaping your web presence so AI assistants describe you accurately and recommend you appropriately. Sometimes called GEO (Generative Engine Optimization). AIO replaces or extends traditional SEO depending on how much of your buyer journey runs through AI.

The technique of giving an AI model a clear, structured signal — typically Schema.org markup on your own website — that it can lean on when answering questions about you. A well-anchored entity is one the model can confidently identify and describe.

Any external reference to your name on the open web. In the AI era, brand mentions are a primary signal AI engines use to decide whether you exist, what you do, and whether to recommend you.

A specific reference to a source within an AI-generated answer. Perplexity is the engine most aggressive about showing citations; ChatGPT and Claude only show them when the user opts in. From a reputation perspective, being cited is the gold standard — it means the AI is willing to attach its claim about you to a source it trusts.

The single 0–10 number that combines the per-engine scores into one figure. MirrorAI weights engines slightly — Perplexity is weighted lower because it historically hallucinates more, while Grok is weighted higher because it tends to honestly admit when it lacks information.

The internal bar an AI model uses to decide whether to answer with specifics or hedge with "I don't have information." Different models tune this differently. Claude is the most cautious; ChatGPT the most willing to fill gaps with plausible-sounding guesses.

A more SEO-flavored synonym for AI Optimization. GEO emphasizes the goal of influencing what a generative engine says when no SERP is shown. The label matters less than the underlying work, which is the same as AIO.

The technique of forcing an AI model to base its answer on retrieved documents rather than relying on its training data. Gemini "grounding" specifically refers to grounding answers in Google Search results.

An AI model generating confident-sounding text that is factually wrong. In a reputation context, hallucinations show up as invented credentials, wrong locations, and services you have never offered. Hallucinations are why "the AI mentioned us" is not the same as "the AI got us right."

The underlying neural network that powers AI assistants like ChatGPT, Gemini, Claude, Perplexity, and Grok. LLMs are trained on massive text datasets and then aligned with human feedback to behave like conversational assistants.

An emerging standard (2024–2025) where a website publishes a structured plain-text file telling LLMs what the site is about, who runs it, and what its key resources are. Think of it as robots.txt for AI assistants. View MirrorAI's llms.txt.

How frequently your name appears across the open web relative to your category. Higher mention density correlates strongly with stronger AI recognition. Quality matters too — mentions on high-authority sites count for much more than mentions on low-authority ones.

The date after which an AI model's training data no longer includes new web content. Anything that happened after the cutoff is invisible to the model unless the model is using live retrieval. Claude has historically had longer effective cutoffs because it does not retrieve by default.

The 0–10 score MirrorAI assigns to a single AI engine for a single subject. Per-engine scores reveal the asymmetries that a composite score hides — for example, being 10/10 on ChatGPT and 2/10 on Perplexity is a very different problem from being uniformly 6/10 across all engines.

The fast first stage of MirrorAI's scoring pipeline that looks for explicit "I have no information" patterns in AI responses and assigns a low score before the heavier LLM-based grader runs. This protects against over-crediting polite non-answers.

The technical pattern where, before answering, an AI system fetches relevant documents from a search index and passes them to the LLM as context. Most modern AI assistants use some form of RAG when accuracy matters more than speed.

One of the four dimensions in the MirrorAI score. Recognition measures whether the AI confidently identifies you or your business by name and category. Low recognition means the AI does not know you exist.

The live web-search step that some AI assistants run before composing an answer. Perplexity is the most retrieval-heavy engine; Gemini has built-in Google retrieval; Claude historically does not retrieve unless explicitly configured to.

A standardized vocabulary for describing things on the web in a way machines can read. Person, Organization, Service, FAQPage, HowTo. Schema.org markup is one of the highest-leverage technical moves you can make to improve how AI tools understand your site.

The classic discipline of ranking a page on Google's results page. Still useful, especially because Gemini grounds answers in Google search, but no longer the whole game. AI Reputation overlaps with SEO but is not identical to it.

The weight an AI model gives to a particular source. Wikipedia is treated as a high-trust source by almost every model. A new self-published blog is treated as low-trust. Source trust explains why one mention in a major publication can outweigh fifty self-published posts.

The periodic process by which AI labs retrain their models on updated data. Training cycles happen on the scale of months. Anything that happens between cycles is invisible to the model unless the model retrieves it at query time.

The plain-English summary MirrorAI attaches to a composite score, such as "Invisible," "Critical," "Moderate," "Strong," or "Authority." The verdict translates a number into an intuition for what buyers actually experience when they ask AI about you.

The degree to which an AI tool can find and describe you. A business with high visibility appears in AI answers consistently across engines. Visibility is upstream of recommendation — the AI cannot recommend what it cannot see.