// reference

AI & Software Glossary

Plain-English definitions of the terms we use every day — from foundational machine learning concepts to the latest AI architectures.

AI Agent: An autonomous software program that perceives its environment, makes decisions, and takes actions to achieve a goal — without requiring step-by-step human instruction. Modern AI agents use large language models as their reasoning core.
AI-First: A development philosophy where artificial intelligence is embedded into the core of a product's architecture from day one, rather than added as an afterthought. AI-first products treat intelligence as a first-class feature.

Computer Vision: A field of AI that enables machines to interpret and understand visual information from images and video. Applications include defect detection, medical imaging, object recognition, and real-time video analytics.

Embeddings: Numerical representations of text, images, or other data in a high-dimensional vector space, where semantically similar items are placed closer together. Embeddings power semantic search and are the foundation of RAG systems.

Fine-tuning: The process of further training a pre-trained model on a smaller, domain-specific dataset so it adapts to a particular task or writing style. Fine-tuning is more cost-effective than training a model from scratch.
Foundation Model: A large AI model trained on broad, general-purpose data that can be adapted to a wide range of downstream tasks. GPT-4, Claude, and Gemini are examples of foundation models.

Generative AI (GenAI): AI systems that can generate new content — text, images, audio, code, or video — by learning patterns from training data. Large language models are the most widely deployed type of generative AI.

Hallucination: When an AI model produces output that is plausible-sounding but factually incorrect or fabricated. Hallucinations are a key risk in production AI systems and are mitigated through techniques like RAG and output validation.

Inference: The process of running a trained AI model on new input data to generate predictions or outputs. Inference is distinct from training — it is what happens when you use an AI model in production.

LLM (Large Language Model): A type of AI model trained on massive text corpora that can understand and generate human language. LLMs form the backbone of modern AI assistants, code generation tools, and document analysis systems.

MLOps: A set of practices that combines machine learning, DevOps, and data engineering to automate and streamline the end-to-end machine learning lifecycle — from data preparation and model training to deployment and monitoring.
Model Drift: The degradation of an AI model's performance over time as the real-world data it encounters diverges from the data it was trained on. Monitoring for drift is a core MLOps practice.
Multimodal AI: AI systems that can process and generate multiple types of data — such as text, images, audio, and video — simultaneously. GPT-4o and Gemini are examples of multimodal models.

Natural Language Processing (NLP): A branch of AI concerned with enabling machines to understand, interpret, and generate human language. NLP underpins applications like sentiment analysis, text summarisation, chatbots, and machine translation.
Neural Network: A computational model loosely inspired by the human brain, composed of layers of interconnected nodes (neurons) that learn patterns from data. Deep neural networks with many layers are the foundation of modern AI.

Predictive Analytics: The use of statistical algorithms and machine learning to forecast future outcomes based on historical data. Common applications include demand forecasting, churn prediction, and risk scoring.
Prompt Engineering: The practice of crafting, structuring, and optimising the inputs given to a language model in order to steer its outputs toward a desired result. Prompt engineering is a critical skill for building reliable AI applications.

RAG (Retrieval-Augmented Generation): An architecture that enhances a language model's responses by first retrieving relevant documents or data from an external knowledge source, then feeding that context into the model alongside the user's query. RAG reduces hallucinations and keeps AI responses grounded in up-to-date facts.

Semantic Search: Search that understands the meaning and intent behind a query, rather than relying purely on keyword matching. Semantic search uses embeddings to find results that are conceptually related, even if they use different words.

Tool Use (Function Calling): A capability that allows a language model to call external tools — such as APIs, databases, or code interpreters — as part of generating a response. Tool use is what transforms a language model into an AI agent that can take real-world actions.

Vector Database: A database optimised for storing and querying high-dimensional vector embeddings. Vector databases power semantic search and RAG systems by enabling fast similarity search across millions of embedded documents.

Let's work together

Have a challenge?
We'll build the solution.

No pitch decks. Just a conversation about what you need.