Reference

AI Glossary

Clear, short definitions of the most important AI terms. Each entry is a standalone, fact-based answer, linked to in-depth articles and relevant services.

AEO (Answer Engine Optimization): AEO is the practice of optimizing content so it is served as the direct answer in answer engines, voice assistants or featured snippets.
AI Agent: An AI agent is a system that uses a language model to autonomously break a goal into steps, call tools or APIs, and complete multi-step tasks without constant human direction.
AI Phone Assistant: An AI phone assistant is a system that automatically handles inbound or outbound phone calls, understands the caller's request in natural language and responds by voice in real time.
Artificial Intelligence (AI): Artificial Intelligence is the field of building computer systems that perform tasks normally requiring human intelligence, such as understanding language, recognising patterns or making decisions.
Automation: Automation is the use of technology to perform tasks or processes with minimal human intervention, replacing manual steps with rule-based or AI-driven procedures.
Chatbot: A chatbot is a software program that holds text-based conversations with users, either rule-based following fixed flows or powered by AI language models.
Conversational AI: Conversational AI refers to technologies that let machines understand natural language in text or speech and respond contextually within a dialogue.
Deep Learning: Deep Learning is a subfield of machine learning that uses artificial neural networks with many layers to learn complex patterns in large amounts of data such as images, text or speech.
E-E-A-T: E-E-A-T stands for Experience, Expertise, Authoritativeness and Trustworthiness — criteria from Google's Search Quality Rater Guidelines used to assess the quality of content.
Embedding (vector): An embedding is a representation of data such as words, sentences or images as a vector of numbers in which semantically similar items lie close together.
EU AI Act: The EU AI Act is the European Union's regulation establishing a risk-based legal framework for artificial intelligence; it entered into force on 1 August 2024 and applies in phases.
Fine-Tuning: Fine-tuning is the targeted further training of an already pretrained AI model on a smaller, task- or domain-specific dataset to adapt its behaviour for a particular purpose.
Generative AI: Generative AI refers to AI systems that create new content such as text, images, audio, video or code by applying patterns learned from training data to new inputs.
GEO (Generative Engine Optimization): GEO is the practice of optimizing content so it gets cited and surfaced within the AI-generated answers of generative engines such as ChatGPT, Perplexity or Google AI Overviews.
Google AI Overviews: Google AI Overviews are AI-generated summaries shown at the top of Google Search results that synthesize information from multiple web sources and include links.
Hallucination (AI): A hallucination is a statement produced by an AI model that sounds fluent and plausible but is factually wrong, fabricated or unsupported by the sources.
Inference (AI): Inference is the process of applying a trained AI model to new input to produce an output such as a prediction or generated text — as opposed to training.
Large Language Model (LLM): A Large Language Model is an AI model trained on very large amounts of text that predicts the next token and thereby understands and generates language.
llms.txt: llms.txt is a proposed Markdown-formatted text file placed at a website's root that gives large language models a curated, machine-friendly overview of the site's key content.
Local AI (on-device / self-hosted): Local AI refers to running AI models on your own device or infrastructure instead of a third-party cloud, so that data stays local.
Machine Learning: Machine Learning is a subfield of AI in which systems learn patterns from data and improve their performance without being explicitly programmed for each task.
Neural Network: A neural network is a computational model of connected nodes ("neurons") arranged in layers that processes inputs through weighted connections and adjusts those weights during training to perform tasks.
nFADP (revised Swiss Federal Act on Data Protection): The nFADP is Switzerland's revised Federal Act on Data Protection, in force since 1 September 2023, governing the processing of personal data in Switzerland.
Open-Source AI: Open-Source AI refers to AI systems, especially models, whose components such as weights, code and sometimes training data are released under licenses permitting use, study, modification and redistribution.
Prompt: A prompt is the input, usually phrased as text, with which a user gives an AI model a task, question or instruction.
Prompt Engineering: Prompt engineering is the practice of deliberately formulating and refining inputs to AI models in order to obtain more reliable, accurate and usable results.
Retrieval-Augmented Generation (RAG): Retrieval-Augmented Generation (RAG) is a method in which a language model retrieves relevant documents from an external knowledge source before answering and grounds its output on them.
SEO (Search Engine Optimization): SEO is the practice of optimizing websites and content to rank higher in organic (unpaid) search-engine results and attract more relevant visitors.
Speech-to-Text (STT): Speech-to-Text (STT), also called speech recognition, is a technology that automatically converts spoken language from an audio signal into written text.
Structured Data (Schema.org): Structured data is machine-readable markup, using the Schema.org vocabulary, added to a web page so search engines can understand its content and display rich results.
Text-to-Speech (TTS): Text-to-Speech (TTS), also called speech synthesis, is a technology that automatically converts written text into spoken, as natural-sounding as possible, speech.
Token (in LLMs): A token is the smallest unit of text a language model processes, usually a word, part of a word or a punctuation mark, into which text is split before processing.
Transformer (architecture): The transformer is a neural network architecture introduced in 2017 that uses an attention mechanism to weigh the relationships between all elements of a sequence, enabling modern language models.
Voice AI: Voice AI is the umbrella term for AI technologies that process and generate spoken language, including speech recognition, speaker identification, language understanding and speech synthesis.
Voicebot: A voicebot is a dialogue system that interacts through spoken language, converting speech to text, interpreting the request and replying via synthesised speech.
Workflow Automation: Workflow automation is the automation of a connected sequence of tasks across a business process, so that information and triggers flow automatically from one step or system to the next.