/ Glossary

AI, automation & generative SEO glossary

TL;DR — Glossary of 18 terms — generative AI, agents, RAG, GEO, AIO, n8n automation — each defined in under 60 words.

An operational glossary of the terms we use in briefs, audits and proposals. No filler jargon — every entry is two or three sentences.

AI Agent

Autonomous program that uses a language model to plan, call tools (APIs, databases, browsers) and execute a multi-step task without human input at each step.

An AI agent differs from a chatbot in its ability to act — it triggers workflows, writes to third-party systems and loops until it hits a goal.

See also: RAG (Retrieval-Augmented Generation), n8n

RAG (Retrieval-Augmented Generation)

Architecture that pairs a vector search engine with an LLM. The system retrieves relevant passages from a private corpus and injects them into the model's context to generate a sourced answer.

RAG solves hallucination on proprietary data without retraining the model. Low marginal cost, instant updates.

See also: Fine-tuning, Embedding

Fine-tuning

Process that adjusts the weights of a pre-trained model on a domain-specific dataset. The model learns a style, format or reasoning pattern — not facts.

See also: RAG (Retrieval-Augmented Generation), LLM (Large Language Model)

Embedding

Numerical representation (a vector of several hundred dimensions) of a piece of text, image or audio. Two semantically close items sit close in vector space.

See also: RAG (Retrieval-Augmented Generation)

LLM (Large Language Model)

Large-scale language model trained on trillions of tokens. Examples: GPT-5, Claude 4, Gemini 2, Llama 3. Generates text, code and calls tools through tool-use.

GEO (Generative Engine Optimization)

Discipline that optimises a site to be cited by generative engines (ChatGPT, Perplexity, Google AI Overviews, Claude). Built on semantic structure, schema.org markup and factual clarity.

See also: AIO (AI Overview), SEO (Search Engine Optimization)

AIO (AI Overview)

AI-generated answer block displayed above Google's organic results. Synthesises multiple sources with links. Earning an AIO citation is now more strategic than ranking #1.

See also: GEO (Generative Engine Optimization), SEO (Search Engine Optimization)

SEO (Search Engine Optimization)

Optimisation for traditional blue-link search engines. Covers technical, content, authority (backlinks) and user experience. Still essential, but no longer sufficient on its own in 2026.

See also: GEO (Generative Engine Optimization), AIO (AI Overview)

n8n

Open-source, self-hostable workflow automation platform. 400+ connectors, inline JavaScript, unlimited executions. An alternative to Zapier and Make that scales without per-task pricing.

See also: AI Agent, Automation

Automation

Replacing repetitive manual tasks with workflows orchestrated across systems (CRM, ERP, email, databases). Common targets: billing, lead routing, tier-1 support, report generation.

See also: n8n, AI Agent

Token

Unit of text processing for an LLM. Roughly 0.75 words in English. API calls are billed per input and output token.

Prompt engineering

Designing and iterating on instructions given to an LLM to get reliable, repeatable outputs. Includes few-shot, chain-of-thought, structured outputs and guardrails.

Vector database

Database specialised in storing and searching embedding vectors by cosine similarity. Examples: Pinecone, Qdrant, pgvector. A foundational building block of any RAG architecture.

See also: Embedding, RAG (Retrieval-Augmented Generation)

MCP (Model Context Protocol)

Open protocol introduced by Anthropic that standardises how an LLM connects to external context sources (files, APIs, databases). Becoming the de-facto standard for agent ecosystems in 2026.

See also: AI Agent

Hallucination

Plausible but factually wrong answer produced by an LLM. Mitigation: RAG on vetted sources, tool-use for numbers, mandatory citations, second-model validation.

See also: RAG (Retrieval-Augmented Generation)

Schema.org

Structured vocabulary (JSON-LD) that describes a page's content for search engines and LLMs. Key types in 2026: Organization, FAQPage, Article, BreadcrumbList, SpeakableSpecification, DefinedTerm.

See also: GEO (Generative Engine Optimization), SEO (Search Engine Optimization)

Speakable

Schema.org property that flags sections of a page suitable to be read aloud by a voice assistant. Prerequisite for appearing in Google Assistant and Alexa answers.

See also: Schema.org

SaaS replacement

Strategy of replacing several per-seat SaaS subscriptions with a custom in-house app built using AI and low-code. Typically pays back in 12 to 24 months for a 30+ person team.