An AI glossary builder solves a now-universal problem: AI vocabulary has spread into every industry faster than the explanations have. Staff hear “hallucination”, “RAG”, and “agent” in meetings without a shared, plain-English definition. This tool builds a prompt that makes an LLM generate a glossary tailored to your field, at the right reading level, with examples your team will recognise.
The terms every professional needs now
Whatever the industry, there are a dozen concepts that now appear in boardrooms, procurement meetings, and product conversations — and a shared definition prevents the confusion that blocks action:
- Hallucination — when a model generates confident-sounding information that is factually wrong. Knowing this term prevents misplaced trust in AI outputs.
- RAG (Retrieval-Augmented Generation) — a technique that gives a model access to a specific set of documents at query time. In healthcare, that might be clinical guidelines; in legal, case law; in retail, product catalogues.
- Agent — an AI system that can take actions autonomously, not just produce text. Increasingly common in operations and customer service.
- Fine-tuning — adapting a general model to a specific task or domain using your own data. In finance, this might mean training on earnings call transcripts; in HR, on job descriptions.
- Context window — the amount of text a model can “hold in mind” at once. Relevant when deciding whether a long document can be processed in one pass.
- Embedding — a numerical representation of text that captures its meaning. The technology behind semantic search (“find contracts similar to this one”).
- Prompt — the instruction you give the model. Understanding this term helps non-technical staff understand why AI outputs vary and how to improve them.
- Token — the basic unit the model processes (roughly a word or word-fragment). Relevant to cost calculations: most API pricing is per token.
How it works
You choose your industry, an audience level (beginner or technical), and a term count. The builder assembles a prompt that lists the core AI terms everyone now encounters and instructs the model to define each one in plain English, illustrate it with an example drawn from your industry, and add any extra terms that matter specifically in your field. Beginner mode favours analogies and bans unexplained jargon; technical mode keeps definitions precise. The prompt is built entirely in your browser — you run it in whatever LLM you prefer, then paste the result into a wiki, deck, or onboarding doc.
Tips and examples
For onboarding non-technical teams, choose beginner and a modest term count (10–15) so the glossary is digestible — you can always generate a second, advanced pass later. Ask the model to follow up with a short quiz to confirm the definitions landed. Industry grounding is what makes this valuable: a “fine-tuning” example for a retailer (“adapting a model to your product catalogue”) lands far better than an abstract one. Keep the glossary in version control or a shared doc and regenerate it as the field moves — terms like “agent” and “MCP” shift meaning quickly, so an annual refresh keeps your team speaking the same language.