Fifty plain-English definitions of the terms that come up in AI consulting conversations with Hawai'i business owners. Each entry is written for the buyer rather than the engineer: short, concrete, with one example. No log-in and no email gate to read them.
Generative Engine Optimization (GEO)
The practice of optimizing a business's online presence so that AI search engines (ChatGPT, Perplexity, Claude, Google AI Overviews, and Gemini) cite that business as a recommended answer when users ask about its services. GEO differs from traditional SEO: SEO ranks you on a list of links; GEO positions you as the answer itself.
Example: A Hawai'i contractor with GEO optimization is named when a customer asks ChatGPT, "who should I call for a hot water heater repair in Honolulu?"
AI Overviews
Google's AI-generated summary that appears above traditional search results for many queries. It is powered by Gemini 3 and pulls from multiple cited sources to construct an answer. The cited sources receive direct traffic when users click through.
Example: Searching "AI receptionist Hawaii" now returns an AI Overview with a written answer and 3-5 citation links before the standard 10 blue links appear.
ChatGPT Search
OpenAI's web-search-enabled feature inside ChatGPT that pulls live web content into its responses with inline citations. It tends to cite fewer sources per query than Perplexity but extracts more language from each one.
Example: When a user asks ChatGPT "compare AI consulting agencies in Hawaii," the response typically cites 5-8 source URLs inline.
Perplexity
A search-first AI engine that retrieves 10+ pages per query and reranks them at the paragraph level, citing the 3-4 most quotable passages. Perplexity has unusually high transparency on which sources it used.
Example: Perplexity's answers display each source as a numbered citation directly tied to the sentence that source informed.
Claude
Anthropic's conversational AI. Claude's web-enabled product uses Brave Search results as its retrieval layer. Claude weights credentialed authors and primary sources heavily in citation selection.
Example: Claude is often the first AI engine to cite a business once it has a complete LinkedIn profile and Wikipedia/Wikidata entry.
Gemini
Google's flagship AI model, powering Google AI Overviews, Gemini app, and Workspace AI features. Gemini 3 (launched Q1 2026) generates 8-12 sub-queries from a single user question, pulling sources for each. That is why a top-10 Google rank no longer guarantees AI citation.
Example: A Gemini-powered AI Overview for "best Maui plumber" may pull from 12 different sources, not just the top-3 Google results.
Query Fan-Out
The process by which an AI search engine expands a single user query into 8-12 related sub-queries, retrieving sources for each. The synthesized answer pulls from the full set, which dilutes the value of ranking #1 on the original query alone.
Example: "AI consulting Hawaii" might fan out into "AI receptionist Maui," "GEO services Honolulu," "Hawai'i AI agency reviews," and 6 more sub-queries.
Schema.org
A shared vocabulary for structured data that websites add to their HTML so search engines and AI systems can understand the page's entities and relationships. Schema.org is the standard backing JSON-LD markup used by Google, Bing, OpenAI, and others.
Example: A Hawai'i AI agency's homepage emits schema.org Organization, LocalBusiness, Service, and FAQPage objects to describe what it is and what it offers.
JSON-LD
JavaScript Object Notation for Linked Data, the recommended format for embedding schema.org structured data in HTML pages. JSON-LD goes inside a <script type="application/ld+json"> tag in the page <head> and is parsed by both search and AI crawlers.
Example: Every page on a GEO-optimized site has at least one JSON-LD script declaring its primary entity (Service, BlogPosting, AboutPage, etc.).
FAQPage Schema
A schema.org type that marks a page as containing question-and-answer pairs. AI engines extract FAQ entries verbatim when answering similar questions. The Q&A content MUST be visible on the page; JSON-LD-only FAQs without visible HTML risk manual penalty.
Example: A service page's FAQ accordion is mirrored exactly into FAQPage schema, so AI engines can cite individual Q&A pairs in answers.
BlogPosting Schema
A schema.org type for individual blog articles. It carries headline, author, datePublished, dateModified, image, and wordCount: all of which AI engines use to weight citation freshness and authority.
Example: A blog post's BlogPosting schema includes a full Person object for the author with a resolving URL to the author's bio page.
Knowledge Graph
Google's database of entities (businesses, people, places, concepts) and their relationships. AI engines use Knowledge Graph IDs to disambiguate "which 808 AI Group" or "which Devin Atkins" they're talking about. Without a knowledge graph entry, you're a text string instead of a known entity.
Example: A Hawai'i business with a Wikidata entry resolves cleanly in Gemini's knowledge graph and is cited more reliably.
Wikidata
An open knowledge graph that anyone can edit (subject to citation requirements). Businesses with a Wikidata entry get a canonical machine-readable identifier (a Q-ID) that AI engines treat as authoritative. Wikidata has no notability threshold for businesses, unlike Wikipedia.
Example: A Hawai'i AI consultancy creates a Wikidata item linking to its LinkedIn page and is then cited 2.7-8.2× more frequently across AI engines.
sameAs Property
A schema.org property that links your entity to its identifiers on other authoritative platforms: LinkedIn, Wikidata, Crunchbase, and official social profiles. AI engines use sameAs to verify that your text-string brand name matches a known entity.
Example: An Organization schema with sameAs pointing to LinkedIn and Wikidata receives substantially higher citation rates than one without external identifiers.
Canonical URL
The official URL for a given piece of content, declared via <link rel="canonical">. When the same content is reachable at multiple URLs (with/without trailing slash, with tracking params), the canonical tells search and AI engines which version to treat as the source.
Example: Every page on the808aigroup.com declares its canonical URL so AI crawlers don't fragment authority across variant URLs.
robots.txt
A plain-text file at the root of a website that tells crawlers which URLs they may or may not request. In 2026, robots.txt also lists AI-specific crawlers (GPTBot, ClaudeBot, PerplexityBot) with explicit allow or disallow rules.
Example: A GEO-optimized robots.txt explicitly allows OAI-SearchBot, PerplexityBot, and Claude-SearchBot, the inference crawlers that drive live citations.
llms.txt
A proposed convention (Answer.AI, September 2024) for sites to provide AI systems with a curated summary of their key pages. Adoption remains limited; most major AI engines do not currently read it, though some developer tools do.
Example: A minimal, well-curated llms.txt lists 5-10 key URLs with one-line descriptions for AI systems that respect the convention.
Prerendering
Generating static HTML for each route of a JavaScript single-page app at build time, so AI crawlers that can't execute JavaScript still see the full page content. Prerendering is the cheapest fix for the "SPA invisible to AI crawlers" problem.
Example: After prerendering, requesting any URL with curl returns complete HTML with content and schema: the same content the React app shows users.
AI Receptionist
A voice-powered AI agent that answers business phone calls 24/7, qualifies callers, answers FAQs, books appointments, and routes complex inquiries to human staff. Modern AI receptionists handle multiple simultaneous calls without hold times.
Example: A $97 AI Receptionist setup at a Hawai'i restaurant answers reservation calls in under two rings, books OpenTable slots automatically, and texts the owner only when a caller needs human escalation.
AI Automation
End-to-end workflow automation that uses AI models for the steps requiring judgment (classification, summarization, extraction) and conventional automation for the deterministic steps (data movement, notifications, integrations). Used to remove manual handoffs in operations like booking, follow-up, invoicing, and reporting.
Example: AI automation for a Hawai'i plumbing company handles dispatch confirmation, customer rescheduling, invoice generation, and review requests without admin staff intervention.
Workflow Systems
The mapping and design of how work moves through an organization (intake, routing, processing, review, output) with AI agents and automation embedded at specific decision points. The method maps the shape of the work first, then identifies where AI fits.
Example: A Hawai'i contractor's workflow system maps lead intake (AI receptionist), qualification (AI booking), estimate generation (AI estimator), and follow-up (automated sequence) as one connected pipeline.
Lead Generation Systems
Multi-channel acquisition that combines paid advertising, organic content, and AI-powered follow-up sequences to deliver qualified leads to a sales pipeline. Modern lead-gen systems use AI for ad copy variation, lead qualification, and personalized re-engagement.
Example: A Maui realtor's lead-gen system runs Facebook ads, captures inquiries through a chatbot, qualifies on intent, and routes hot leads to the agent's phone in real time.
Funnel Design
The structured progression a prospect moves through from first awareness to becoming a customer: typically lead magnet → low-ticket offer → recurring program → high-ticket consultation. Good funnel design matches each stage's offer to the buyer's level of commitment.
Example: A four-tier revenue architecture: free GEO audit (lead magnet) → $97 AI receptionist (low ticket) → monthly GEO retainer (recurring) → custom app build (high ticket).
AI Voice Agent
A specific class of AI assistant designed to handle voice interactions over phone calls. Voice agents combine speech-to-text, language understanding, response generation, and text-to-speech in low-latency pipelines that feel conversational rather than robotic.
Example: A modern AI voice agent answers a Hawai'i restaurant's phone, understands a reservation request in Pidgin or English, books the table in OpenTable, and confirms, typically in under 90 seconds.
Conversational AI
AI systems designed to maintain coherent, contextual dialogue across multiple turns. Conversational AI powers chatbots, voice assistants, and AI receptionists. Quality depends on how well the system handles ambiguity, memory of prior turns, and graceful handoff to humans.
Example: A conversational AI on a Hawai'i contractor website handles "do you do plumbing in Lahaina?": it recognizes the location, confirms the service area, and books an estimate.
Large Language Model (LLM)
A neural network trained on vast amounts of text that can understand and generate human language. LLMs like GPT-4, Claude 3.7, and Gemini Ultra are the engines behind modern AI assistants and AI search.
Example: An AI receptionist uses an LLM under the hood to interpret what a caller is asking and generate an appropriate response in real time.
RAG (Retrieval-Augmented Generation)
A technique where an AI system retrieves relevant documents from a knowledge base before generating a response, combining the language fluency of LLMs with the factual accuracy of search. RAG underpins most enterprise AI assistants and AI search engines.
Example: A Hawai'i medical office's internal AI assistant uses RAG to answer staff questions by retrieving from the office's policy documents before composing the answer.
Prompt Engineering
The craft of structuring instructions, examples, and context to get reliable, useful outputs from an LLM. Better prompts produce better answers; weak prompts produce vague or wrong answers from the same underlying model.
Example: A Hawai'i agency's prompt for its blog-drafting AI specifies tone, length, structure, banned phrases, and required local references, which produces far more usable drafts than a generic prompt.
AI Hallucination
When an AI model generates information that sounds plausible but is factually wrong or fabricated. Hallucinations are highest risk in fact-dense domains (medical, legal, financial) and lowest in well-grounded retrieval-augmented systems.
Example: Without retrieval-augmentation, an AI receptionist asked "what time do you close?" may hallucinate hours; with RAG against the actual business profile, it answers correctly.
Bottom Line Up Front (BLUF)
A writing structure that puts the direct answer in the first paragraph, before context or backstory. AI engines extract the first 30% of a page disproportionately (per Princeton GEO research), which makes BLUF structure mechanically advantageous for citation.
Example: A service page's BLUF paragraph names the business, location, and core service in the first 50 words instead of opening with brand storytelling.
Citation Lift
The measurable increase in how often a business is cited by AI search engines, typically tracked over a 30/60/90-day window after an optimization. A well-executed GEO program targets meaningful citation lift within 60 days.
Example: Baseline AI citations for "AI consulting Hawaii" were 0; 60 days after implementation, the business is cited in 5 of 10 sample queries across ChatGPT, Claude, and Perplexity.
Passage-Level Retrieval
The mechanism by which AI engines select individual paragraphs (not whole pages) as citation units. Perplexity in particular ranks paragraphs as standalone retrieval objects, so a single high-quality paragraph on a weak page can outrank a weak paragraph on a strong page.
Example: A 180-word paragraph that defines a concept, cites a statistic, and includes a concrete example is the unit AI engines extract, not the page it lives on.
Semantic Completeness
The degree to which a page's content fully answers the question a user asked, including obvious follow-up questions. AI engines prefer pages that don't require the user to click elsewhere for the rest of the answer.
Example: A semantically complete service page covers what the service is, who it's for, what it costs, how long it takes, who provides it, and what the next step is, all on one page.
Long-Tail Query
A specific, multi-word search query with relatively low individual search volume but collectively significant total volume. Long-tail queries are easier to rank for and convert at higher rates because they signal specific intent.
Example: "AI receptionist for Hawai'i restaurant under $200" is a long-tail query: low volume but very high intent.
Click-Through Rate (CTR)
The percentage of impressions on a search result that produce a click. CTR is influenced by ranking position, title quality, meta description, and presence of rich results (stars, FAQ snippets, sitelinks).
Example: Improving a page's title and meta description can raise CTR from 2% to 6% without any change in actual ranking position.
Bounce Rate
The percentage of users who arrive on a page and leave without further interaction. High bounce rate often signals weak match between query and content. AI engines weight engagement signals as a quality proxy.
Example: A Hawai'i service page with a 70% bounce rate from "AI receptionist Maui" suggests the page didn't deliver what the searcher expected to find.
Local Pack
Google's three-result map and listing block that appears for local-intent queries above traditional organic results. Inclusion is driven by Google Business Profile completeness, review velocity, citation consistency, and proximity to the searcher.
Example: Ranking in the Google Local Pack for "plumber Honolulu" delivers more calls than ranking #1 in traditional organic results below the pack.
Google Business Profile (GBP)
Google's free business listing tool, formerly Google My Business. A complete, verified, actively-managed GBP drives Local Pack rankings, AI-engine local citations, and direct calls. AI engines explicitly reference GBP data in local-business answers.
Example: A Hawai'i business with 150+ GBP reviews and consistent posts is materially more likely to be cited in AI Overviews than one with 12 reviews.
NAP Consistency
The exact match of business Name, Address, and Phone across every web listing: website, GBP, directories, social profiles, and press mentions. Inconsistent NAP data fragments entity recognition for both Google and AI engines.
Example: If your website says "+1 (808) 490-6238" but your Clutch profile says "808-490-6238," entity resolution downgrades, even though both are the same number.
Citation (local SEO)
A mention of your business name, address, or phone on another website: in directories, blogs, news, or social. Citation count and consistency are foundational local SEO signals. Distinct from AI citations.
Example: Listings on Yelp, Yellow Pages, Chamber of Commerce Hawaii, and the local newspaper all count as NAP citations that reinforce local authority.
Service Area
The geographic region a business serves, declared in Google Business Profile and in schema.org LocalBusiness markup. Accurate service area declaration is essential for inter-island businesses serving multiple Hawaiian islands.
Example: A Maui-based agency declares service areas including Maui, O'ahu, Big Island, and Kaua'i, with island-level disambiguation in schema.
E-E-A-T
Google's quality framework (Experience, Expertise, Authoritativeness, Trustworthiness) used to evaluate content quality. AI engines apply similar criteria when selecting citations: bylined authors, credentials, primary-source data, and external recognition all contribute.
Example: A blog post written by a named founder with a linked About page and a citation to a primary research paper outperforms an anonymous post on the same topic.
Comparison Page
A page that directly compares two options ("X vs Y") with structured analysis: typically a comparison table, pros and cons, and a recommendation. AI engines cite comparison pages at 61% rate per PresenceAI 2026 research.
Example: "AI Receptionist vs Traditional Answering Service" is a comparison page that ranks for high-intent decision queries.
Case Study
A detailed account of how a business solved a specific client's problem with quantified outcomes. Case studies with real numbers outperform testimonials by ~300% in AI citation rates (Hashmeta 2025).
Example: A case study showing exactly how much a restaurant recovered in missed bookings after deploying an AI receptionist, with the real numbers, is citable; a generic testimonial is not.
Glossary
A structured collection of term definitions on a single page or set of pages, marked with DefinedTermSet schema. Glossaries are unusually high-leverage GEO assets because each entry is a self-contained answer to a "what is X" query.
Example: This page is a glossary: 50 self-contained definitions, each schema-marked, each a potential citation unit for AI search.
CMMC 2.0
Cybersecurity Maturity Model Certification 2.0, the Department of Defense's compliance framework for contractors handling controlled unclassified information. AI tools deployed in CMMC environments must satisfy specific data-handling and audit requirements.
Example: A Hawai'i defense contractor's AI workflow review verifies that no controlled data passes through consumer LLM APIs and that human-in-the-loop checkpoints exist where required.
NIST 800-171
The National Institute of Standards and Technology publication specifying security requirements for protecting Controlled Unclassified Information. The foundation under CMMC and most federal contracting cybersecurity expectations.
Example: AI tools used by Hawai'i federal contractors are evaluated against NIST 800-171 controls before deployment, particularly around access control, audit logging, and incident response.
Transient Vacation Rental (TVR)
A property zoned and permitted for short-term (under 30 days) rental in Maui County, with specific operational conditions. TVR compliance under the County's Bill 9 phase-out is one of the most demanding STR regulatory regimes in the United States.
Example: An AI-driven STR compliance system on Maui validates guest stays against TVR permit conditions before bookings are accepted.