Glossary¶

A glossary of common terms using throughout this project.

Agent

A system where an AI model is given tools, memory, or the ability to take actions, and operates with some degree of autonomy to complete a goal.

Deep learning

A subset of machine learning that uses neural networks with many layers to approach complex tasks.

Guardrail

A constraint applied to a model's behaviour to prevent unwanted outputs (e.g. harmful content, going off-topic, hallucinating facts).

LLM - Large Language Model

A type of AI model trained on large amounts of text data that can generate, summarise, and reason about language.

Local model

An AI system run on your own machine rather than accessed via the cloud.

Machine learning

A subset of AI where systems learn patterns from data rather than following explicitly programmed rules. Deep learning and LLMs fall within this category.

Prompt

The input given to an AI that tells it what to do.

SDD - Spec-Driven Development

A software development methodology where behaviour is defined in a formal specification document first, and code is then generated and validated against that spec.

Token

The unit a language model processes text in. Models have a context window measured in tokens.

If you would like to add a term to the glossary, please file an issue or open a pull request on GitHub.