Resources
The enterprise AI glossary.
Plain-language definitions of the terms that come up when you take AI from an idea to something running in production. No jargon for its own sake, and no vendor spin.
A
- Agent (AI agent)
- An AI system that pursues a goal across multiple steps, calling tools, querying systems, and making decisions, rather than only answering a single prompt.
- Agile
- A way of working that delivers value in small, fast, iterative cycles and adapts as teams learn, instead of planning everything up front.
- AI readiness
- How prepared an organization's data, technology, talent, strategy, process, and governance are to support AI in production.
- Anomaly detection
- A machine learning technique that flags data points or events deviating from normal patterns. Used for fraud, security, and quality monitoring.
C
- Change management
- The structured practice of helping people adopt new ways of working, so a new capability actually gets used.
- Classification
- A machine learning task that assigns inputs to predefined categories, such as labeling a transaction as fraudulent or legitimate.
- Cloud-native
- Software designed to run on modern cloud infrastructure, built to scale elastically and recover from failure by design.
- Copilot
- An AI assistant embedded in a tool or workflow that drafts, suggests, or automates work alongside a human user.
D
- Data engineering
- The discipline of building the pipelines, storage, and governance that turn raw data into trusted, ready-to-use assets.
- Data lakehouse
- A platform that combines the flexible, low-cost storage of a data lake with the structure and performance of a data warehouse.
- Data pipeline
- An automated flow that moves and transforms data from source systems into a usable destination.
- DevOps
- Practices that unite software development and operations to deliver software continuously and reliably.
E
- Embeddings
- Numerical representations of text or other data that capture meaning, enabling search, similarity, and retrieval.
- ETL and ELT
- Patterns for moving data, Extract Transform Load or Extract Load Transform, that prepare raw data for analytics and AI.
F
- Feature engineering
- Turning raw data into the input signals a model needs to make accurate predictions.
- Fine-tuning
- Adapting a pre-trained model to a specific task or domain by training it further on targeted data.
- Forward deployed engineering
- A delivery model where small, senior, cross-functional teams embed directly with a client's business and technology people to build and ship solutions fast.
G
- Generative AI
- AI that creates new content, including text, code, images, and audio, typically using large language models.
- Governance, AI
- The policies, controls, and oversight that keep AI secure, compliant, fair, and accountable.
H
- Hallucination
- When a generative model produces confident output that is factually wrong or unsupported. Reduced through grounding, retrieval, and evaluation.
- Human-in-the-loop
- A design where a person reviews or approves AI output before it is acted on, used where the stakes are high.
I
- Inference
- Running a trained model to produce a prediction or output on new data.
- Intelligent automation
- Combining process redesign, RPA, workflow orchestration, and AI to automate end-to-end business processes.
L
- Large language model (LLM)
- A model trained on vast text data that understands and generates human-like language. The engine behind most generative AI.
- Lean startup
- A method for building products through rapid experimentation, validated learning, and iterative release.
M
- Machine learning (ML)
- A branch of AI in which systems learn patterns from data to make predictions or decisions, rather than following explicit rules.
- Medallion architecture
- A data design that refines data through Bronze (raw), Silver (cleaned), and Gold (curated) layers.
- MLOps
- Deploying, monitoring, and retraining machine learning models reliably in production. DevOps applied to ML.
- Model card
- Documentation describing a model's purpose, performance, limitations, and intended use.
- Model drift
- The gradual decline in a model's accuracy as real-world data shifts over time, requiring monitoring and retraining.
N
- Natural language processing (NLP)
- The field of AI focused on understanding and generating human language.
P
- Predictive analytics
- Using historical data and models to forecast future outcomes such as demand, churn, or risk.
- Prompt engineering
- Designing the inputs to a generative model to get reliable, useful output.
- Propensity model
- A model that predicts the likelihood of a specific behavior, for example whether a customer will pay, churn, or buy.
R
- RAG (retrieval-augmented generation)
- A technique that grounds a language model's answers in your own trusted documents and data, improving accuracy and reducing hallucination.
- Responsible AI
- Building AI that is fair, transparent, secure, and accountable.
- RPA (robotic process automation)
- Software bots that automate repetitive, rules-based tasks by mimicking human actions across systems.
S
- SAFe (Scaled Agile Framework)
- A framework for applying agile practices across many teams in a large organization.
- Semantic layer
- A consistent, business-friendly definition of metrics and data that keeps every report and dashboard in agreement.
T
- Technical program management (TPM)
- Coordinating complex, cross-functional technical programs, including scope, dependencies, risk, and delivery, to keep them on track.
V
- Vector database
- A database that stores embeddings and enables fast similarity search. A common building block for RAG and AI agents.
Next step
Ready to put these into practice?
Tell us where you are with AI and we will tell you what is worth doing first, and what is not worth doing at all.
Book a consultation