# AI and data glossary

> Plain definitions of 26 AI and data terms, from MLOps, MLaaS and RAG to price elasticity and intelligent document processing.

URL: https://preview2.snowfox-ai.com/glossary
Last updated: 2026-11-20

## AI agent

A language-model system that carries out a task in steps, calling tools and business systems such as ERP or CRM, rather than only answering a question.

## AI readiness assessment

A review of whether an organization can put AI to productive use. Snowfox reviews five dimensions together: data, technology, strategy, people and culture, and governance and processes.

## Big data

Data whose volume or speed exceeds what a single database handles comfortably, so it needs distributed processing such as Apache Spark, NoSQL stores or cloud warehouses, often in real time.

## Causal machine learning

Models built to estimate what an action will cause, such as the effect of a price change on demand, rather than only what tends to happen alongside it.

## Computer vision

Models that interpret images and video: recognition, object detection, segmentation and video analysis, applied to inspection, document processing and security.

## Cross-price elasticity and cannibalization

How a price change on one product moves the sales of others. Cannibalization is the costly case: a promotion that grows one product by taking sales from another in the same portfolio.

## Customer and offer segmentation

Grouping customers by behavior so each group gets the offer most relevant to it. Machine learning finds groups that conventional segmentation misses.

## Data engineering

Building and maintaining the pipelines, warehouses and analysis layers that make data available and trustworthy for reports and models.

## Data governance

The rules for who owns data, who may use it, how long it is kept and how it is protected: security, retention and privacy you can defend.

## Data pipeline and ETL

An automated sequence that moves data from source systems to where it is used. ETL (extract, transform, load) is the classic form: pull the data, clean and reshape it, load it into a warehouse.

## Data strategy

A statement of which decisions data should improve and what architecture, governance and tooling are needed to make that possible.

## Data warehouse and data mart

A data warehouse is a central store of cleaned, organized data built for analysis. A data mart is a smaller slice of it shaped for one subject or team.

## Entity and relationship extraction

Finding the named things in a text (people, companies, dates, amounts) and how they relate, for example who grants which powers to whom in a contract. Relationships are the harder part.

## Exit-intent

A message or offer shown at the moment a visitor is about to leave a website, used to recover the visit.

## Generative AI

Models that produce text, code, images or audio. In business systems it usually means large language models used for assistants, agents, document work and code migration.

## Intelligent document processing (IDP)

A pipeline that turns documents such as PDF contracts into structured data: reading the page, understanding its layout, extracting entities and relationships, and validating the result.

## Large language model (LLM)

A model trained on very large amounts of text that can read and write natural language and code. It is the engine behind most generative AI systems.

## Machine learning (ML)

Software that learns patterns from data instead of following hand-written rules, and uses them to predict, classify or recommend.

## MLaaS

Machine learning as a service: using pre-built models that a provider hosts and maintains, paid as an operating cost, instead of building and owning the models.

## MLOps

Machine learning operations: the practices for deploying, monitoring and improving machine-learning models in production with a repeatable process. It is to models what DevOps is to software.

## Model drift

The gradual loss of accuracy of a model in production as customer behavior, prices or products move away from the data it was trained on. Monitoring and retraining are the remedy.

## Natural language processing (NLP)

Techniques that let software read, classify and produce text and speech: classification, entity extraction, summarization, generation and translation.

## Price elasticity

How much the quantity sold of a product changes when its price changes. An elasticity of -2 means a 1% price increase reduces units sold by about 2%.

## Reinforcement learning

A way of training a system by trial and feedback: it tries options, observes the result and shifts toward what works. Used to keep improving which offer or creative is shown.

## Remote sensing

Measuring land and crops from a distance with satellite or drone imagery, used in agriculture to map cultivated areas, monitor crop health and forecast yields.

## Retrieval-augmented generation (RAG)

A design in which the system first retrieves relevant passages from your own documents and then has a language model answer from them. It is how enterprise search and internal assistants stay grounded in company data.
