PREECURSOR
Glossary

The applied-AI glossary

The terms that come up when AI meets production — defined the way we'd explain them to a client, not the way a textbook would.

A–Z
Termagentic AIAgentic AI is the broad class of systems that pursue goals autonomously — planning, calling tools, and adapting over multiple steps instead of answering a single prompt.
TermAI agentsAn AI agent is a system that uses a language model to decide and take actions toward a goal — calling tools, observing results, and looping until the task is done.
TermAI governanceAI governance is the framework of policies, controls, and accountability that decides how an organization builds, approves, monitors, and retires AI systems — so the technology is used responsibly and defensibly.
Termcontext windowA context window is the maximum amount of text a model can take in for a single request — its working memory, measured in tokens, for everything you send and everything it generates.
TermembeddingsEmbeddings are numerical vectors that capture the meaning of text or other data, so that items with similar meaning sit close together in the vector space.
TermevalsEvals are structured tests that measure how well an AI system performs against a set of real tasks, turning model quality from a feeling into a number you can track.
Termfine-tuningFine-tuning continues training a pretrained model on your own examples, adapting its weights so it reliably handles a specific task, format, or domain.
TermguardrailsGuardrails are the controls placed around an AI system that constrain what it can say or do — checking inputs and outputs and bounding the actions it's allowed to take.
TermhallucinationA hallucination is when an AI model produces content that is fluent and confident but factually wrong or unsupported by any source.
TerminferenceInference is the act of running a trained model to generate output from an input — the live operation you pay for and wait on every time an AI system responds.
Termlarge language modelA large language model (LLM) is an AI system trained on vast amounts of text to predict the next piece of text, which lets it generate, summarise, classify, and reason over language.
TermMLOpsMLOps is the discipline of deploying, monitoring, and maintaining machine learning and AI systems reliably in production — the operational backbone that keeps models working after launch.
TermModel Context Protocol (MCP)The Model Context Protocol (MCP) is an open standard that gives language models and agents a uniform way to connect to external tools and data sources, so each integration is built once and reused across applications.
Termmultimodal AIMultimodal AI describes models that take in or produce more than one kind of data — text, images, audio, video — in a single system, rather than handling only text.
Termprompt engineeringPrompt engineering is the practice of designing the instructions, examples, and context you give a language model to get reliable, accurate, well-structured outputs.
Termretrieval-augmented generation (RAG)Retrieval-augmented generation (RAG) is a pattern that fetches relevant documents at query time and feeds them to a language model, so its answers are grounded in your own data rather than only its training.
TermRLHFRLHF — reinforcement learning from human feedback — is a training technique that uses human preferences to steer a model toward responses people actually find helpful and appropriate.
Termvector databaseA vector database stores embeddings and quickly finds the ones most similar to a query — the retrieval engine that powers semantic search and RAG.

Past the definitions, into production

Knowing the terms is the easy part. We build the systems behind them — and prove they moved the number you care about.

Explore AI consulting