AI agent
A system that uses a model, instructions, tools, and context to take steps toward a goal rather than simply return one response.
Plain-English AI
Useful language for useful decisions.
A practical glossary for leaders and teams building, buying, or operating AI products.
Core terms
Clear enough to use in the next meeting.
The right technical term can help, but only when everyone understands its impact on the product and its users.
Evaluation
A repeatable way to test whether an AI system performs well on representative tasks, quality criteria, and failure cases.
Generative AI
AI that creates new content such as text, images, code, audio, or structured outputs from patterns learned during training.
Human in the loop
A deliberate point where a person reviews, corrects, approves, or takes over a system decision.
Large language model
A model trained on large amounts of text to understand and generate language, commonly used for drafting, analysis, and conversation.
RAG
Retrieval-augmented generation: a pattern that retrieves relevant source information before asking a model to generate an answer.
Prompt
The instructions, context, examples, and inputs provided to a model to shape how it responds to a task.
Vector database
A database designed to retrieve information by semantic similarity, often used to find relevant context for an AI system.
Model drift
A change in system behavior or quality over time as inputs, user behavior, data, providers, or models change.
Make it practical
Turn an AI term into a product decision.
Use the glossary as a starting point, then bring us the workflow and constraint behind the question.