Glossary

Artificial intelligence

LLM (Large Language Model): The AI model family behind ChatGPT, Claude and Gemini: trained on text, it reads and writes language.
RAG (Retrieval-Augmented Generation): The technique that lets an AI answer from your documents instead of its memory.
AI agent: An AI system that doesn't just answer — it takes actions in your tools to complete a task.
Prompt engineering: Writing instructions for AI models so they produce reliable, repeatable results.
Fine-tuning: Re-training an existing AI model on your examples to specialise its behaviour.
Hallucination (AI): When an AI model states something false with complete confidence.
Token (AI): The unit AI models read, write and bill by — roughly three quarters of a word.
Vector database: A database that finds documents by meaning rather than by keyword.

GEO (Generative Engine Optimization): Optimising a site to be cited by AI engines like ChatGPT and Perplexity, not just ranked by Google.
AEO (Answer Engine Optimization): Structuring content so it can be lifted whole as the answer to a question.
llms.txt: A plain-text file at a site's root that tells AI crawlers what the site is and where the substance lives.
IndexNow: A protocol that tells search engines about new pages in hours instead of waiting weeks for a crawl.
Structured data (schema.org): Machine-readable labels in a page's code that tell engines exactly what the content is.
hreflang: The tag that tells engines which language version of a page to show which visitor.
Core Web Vitals: Google's three user-experience metrics: loading speed, interactivity and visual stability.
E-E-A-T: Experience, Expertise, Authoritativeness, Trust — how Google judges who deserves to rank.
Backlink: A link from another website to yours — still the strongest third-party vote of confidence.

MVP (Minimum Viable Product): The smallest version of a product that delivers real value to real users.
PWA (Progressive Web App): A website that installs and behaves like an app — no app store required.
SPA (Single-Page Application): A web app that loads once and then updates in place, like a desktop application.
Prerendering: Generating each page's finished HTML at build time, so crawlers and first paints need no JavaScript.
Headless CMS: A content system that stores and serves content by API, leaving the display to any front end.
API: The contract that lets two software systems talk to each other automatically.
Webhook: A notification one system sends another the instant something happens — the push to an API's pull.
Technical debt: The accumulated cost of past shortcuts in software — paid back with interest in slowness and bugs.

CRM: The system that holds every customer, deal and interaction — the commercial memory of a company.
ERP: The system that runs a company's operations: stock, orders, invoicing, accounting, in one place.
SaaS: Software rented as a subscription and used in the browser — no servers, no installation.
Business process automation: Making software do the repetitive work people currently do by hand between systems.

Web3: Applications built on blockchains: ownership and transactions verified by the network, not one company.
Smart contract: A program on a blockchain that executes agreements automatically, exactly as written.