AI glossary for businesses
AI terms you often meet when a business starts adopting AI, explained in plain language.
- AI (artificial intelligence)
- Software that can do things that used to require human thinking: reading and understanding, writing, recognising images, predicting.
- Generative AI
- The kind of AI that creates new content such as text, images, audio or code based on a user's request.
- Large language model (LLM)
- An AI model trained on a very large amount of text to understand and write natural language. It is the “brain” behind AI chatbots.
- Prompt
- The brief you give the AI. The clearer the context, the task and the format you want, the more usable the result.
- Token
- The small unit of text a model processes. The cost of using AI through an API is usually based on the number of tokens sent in and received back.
- Context window
- The maximum amount of text a model can “remember” in one exchange. Go beyond it and the earliest part is forgotten.
- Hallucination
- When AI gives a wrong answer that sounds very confident. This is why important work needs a human check.
- RAG
- A way to let AI look up your company's documents before answering, so the answer follows the documents instead of guessing. See What is RAG.
- Knowledge base
- A set of documents organised for AI to look up: price lists, policies, guides, FAQs.
- Embedding
- A way of turning text into a series of numbers so a machine can compare how similar two pieces are in meaning. Used for document search in RAG.
- Chatbot AI
- A program that talks with users in natural language, usually to answer questions or collect information.
- AI Agent
- An AI that doesn't just answer but carries out several steps on its own to finish a task, such as looking things up, filling in forms or sending emails, within the permissions it has been given.
- Workflow automation
- Connecting software so that work runs by itself through preset steps whenever an event occurs.
- API
- A gateway that lets two pieces of software exchange data. APIs are what allow AI to be plugged into a website, a CRM or accounting software.
- Fine-tuning
- Further training of an existing model on your own data so it does one specific job better. Most small businesses don't need this step yet.
- Human in the loop
- Designing a process so that a person always reviews the important steps before the AI acts.
- Open-source model
- A publicly released AI model that a business can install on its own server, giving better control over data but requiring someone to operate it.
- Personal data
- Information tied to a specific person, such as full name, phone number or email. Think carefully before putting it into external AI tools.
- POC (proof of concept)
- A small-scale trial to check whether an idea is feasible before making a large investment.
- ROI
- Return on investment: the benefit gained compared with the cost spent. For AI projects, measure it in time saved or extra work handled.