AI terms, in plain language
Short definitions a manager can use in a meeting, each with an everyday example from a business like yours.
The basics
AI agent
An AI agent is software that uses a language model to work towards a goal, deciding which steps to take and which tools to use, instead of only answering one question at a time.
AI automation
AI automation is the use of artificial intelligence to complete business tasks that used to need a person to read, decide or write, such as sorting emails, drafting replies or extracting details from documents.
API (application programming interface)
An API (application programming interface) is a defined way for one piece of software to request information or actions from another, so that systems can work together without a person in between.
Agentic AI
Agentic AI is an approach in which AI systems plan and carry out multi-step work with a degree of independence, choosing actions, using tools and checking their own progress towards a goal.
Chatbot
A chatbot is a software program that holds a text conversation with people, answering questions or guiding them through a task on a website, in an app or on a messaging channel.
Context window
A context window is the amount of text an AI model can take into account at one time, including your instructions, any documents you share and the conversation so far.
Conversational AI
Conversational AI is the group of technologies that lets software understand what people say or type in everyday language and reply naturally, across chat, messaging apps and phone calls.
Generative AI
Generative AI is artificial intelligence that creates new content, such as text, images, audio, video or code, in response to instructions, based on patterns learned from large amounts of existing material.
Intent detection
Intent detection is the step where an AI system works out what a customer wants from their message, such as booking, asking a price or making a complaint. The system then uses that label to route the message to the right answer, workflow or person.
Large language model (LLM)
A large language model (LLM) is an AI system trained on very large amounts of text so that it can understand instructions and produce written answers, summaries, translations and other language tasks.
Multimodal AI
Multimodal AI is artificial intelligence that can work with more than one type of content, such as text, images, audio and documents, understanding them together and sometimes producing them as well.
Prompt
A prompt is the instruction or question you give to an AI model, together with any background, examples or documents you include, which the model uses to produce its answer.
Token
A token is the small unit of text that an AI model reads and writes, often a short word or part of a word, and it is the unit used to measure usage and limits.
Building blocks
AI pilot (proof of concept)
An AI pilot is a small, time-limited trial of an AI tool or agent on one real task, with real data and a few real users, before a business commits to a full rollout. It gives the business evidence about whether the tool works in its own setting.
Embedding
An embedding is a list of numbers that represents the meaning of a piece of text, so that a computer can compare passages and find the ones that are about the same thing.
Fine-tuning
Fine-tuning is the process of further training an existing AI model on a set of your own examples so that it adopts a particular style, format or specialised behaviour more reliably.
Knowledge base
A knowledge base is an organised collection of a company's own information, such as policies, product details, prices and answers to common questions, that people and AI assistants use as their source of truth.
Model Context Protocol (MCP)
Model Context Protocol (MCP) is an open standard that defines how AI assistants connect to external tools and data sources, so one connector can work across different AI applications.
No-code automation
No-code automation is the building of automated workflows with visual tools, by connecting ready-made blocks for apps and actions, without writing programming code.
Prompt engineering
Prompt engineering is the practice of writing, testing and improving the instructions given to an AI model so that it produces accurate, consistent and useful results for a specific task.
Retrieval-augmented generation (RAG)
Retrieval-augmented generation (RAG) is a method where an AI system first searches your own documents for relevant passages and then uses them to write its answer, so replies are grounded in your information.
Skill file
A skill file is a reusable set of written instructions, usually a plain text or markdown document, that teaches an AI assistant how to perform one specific job the way a business wants it done.
Speech-to-text (STT)
Speech-to-text (STT) is technology that converts spoken audio into written text, allowing software to transcribe calls, meetings and voice notes or to understand what a caller is saying.
Structured output
Structured output is an AI model's answer returned in a fixed, predictable format with named fields, such as a form or a JSON record, so that other software can use it directly.
System prompt
A system prompt is a set of standing instructions given to an AI model before any conversation starts, defining its role, tone, knowledge limits and rules for every reply that follows.
Text-to-speech (TTS)
Text-to-speech (TTS) is technology that turns written text into spoken audio using a synthetic voice, used for phone assistants, audio versions of content and spoken notifications.
Tool use
Tool use is the ability of an AI model to call external functions or systems, such as a calendar, a database or a payment link generator, to look up information or carry out an action.
Vector database
A vector database is a type of database that stores embeddings and quickly finds the entries closest in meaning to a question, which lets AI systems search large document collections by meaning.
Webhook
A webhook is an automatic message that one system sends to another the moment something happens, such as a new order or incoming chat, so the second system can react immediately.
Running it in a business
Answer engine optimization (AEO)
Answer engine optimization (AEO) is the practice of writing and structuring a website's content so that AI assistants and search features that give direct answers can find, understand and quote it.
CRM integration
CRM integration is the connection of a customer relationship management system to other tools, such as AI assistants, messaging channels and websites, so that customer information is shared automatically.
Intelligent document processing (IDP)
Intelligent document processing (IDP) uses AI to read documents such as invoices, contracts and ID scans, pull out the fields a business needs, and pass them to other systems without manual typing. Unlike basic OCR, it understands what each field means, not just what text is on the page.
Lead qualification
Lead qualification is the process of finding out whether a new enquiry is a good fit and ready to buy, by collecting key details such as need, budget, timing and authority.
Robotic process automation (RPA)
Robotic process automation (RPA) is software that imitates the clicks and keystrokes a person makes on a screen, in order to repeat rule-based tasks in applications that cannot be connected directly.
Voice agent
A voice agent is an AI system that holds spoken conversations over the phone or through a speaker, understanding the caller, replying in a natural voice and completing tasks such as bookings.
WhatsApp Business API
WhatsApp Business API is the official interface from Meta that lets companies connect WhatsApp to their own software, so they can automate conversations, use AI assistants and serve many customers from one number.
Workflow automation
Workflow automation is the use of software to move a task through its steps automatically, passing information between people and systems according to rules instead of manual copying and reminders.
Safety and governance
AI governance
AI governance is the set of policies, roles and checks an organisation uses to decide how AI may be used, who is accountable for it and how its risks are managed.
Data residency
Data residency is the question of which country or region a company's data is physically stored and processed in, and which laws and contractual terms therefore apply to it.
Evaluation
Evaluation is the practice of testing an AI system against a set of realistic examples and clear criteria, to measure how often it gives correct, safe and useful results.
Guardrails
Guardrails are the rules and technical checks placed around an AI system to keep its behaviour within safe and acceptable limits, such as blocking certain topics, actions or disclosures.
Hallucination
A hallucination is an answer from an AI model that sounds confident and plausible but is factually wrong or invented, such as a made-up price, policy, quotation or reference.
Human in the loop
Human in the loop is a way of designing AI systems so that a person reviews, approves or corrects the AI's work at defined points before it has a real-world effect.
The daily skill, in your inbox
One email each morning with the new drop. Unsubscribe in one click.
