Glossary
AI and operations terms, defined honestly
What each term means, why it matters, and what people commonly get wrong about it. Written to be useful to someone in a budget meeting, not to demonstrate vocabulary.
D
- DevOpsdev ops, platform engineering
- Making the path from a developer's commit to running software fast, automated and reversible — and treating that path as a product with an owner.
- Driftconfiguration drift, model drift
- The gap that opens between how a system is defined and how it actually is — in infrastructure, because someone changed a console; in AI, because the inputs changed.
E
- Evaluation Seteval set, evals, golden dataset
- A collection of real inputs with known-correct outputs, used to score an AI system objectively and catch regressions before they reach users.
G
- GuardrailsAI guardrails, safety controls
- The controls around an AI system that bound what it can do and say — confidence thresholds, scoped tool permissions, input and output filtering, and audit logging.
H
- Hallucinationconfabulation, model hallucination
- When a language model produces something fluent, plausible and false — usually because it was asked a question it had no grounds to answer.
I
- Infrastructure as CodeIaC, Terraform
- Defining servers, networks and databases in version-controlled files instead of clicking through a cloud console, so environments can be reproduced exactly.
L
- LLMOpsMLOps for LLMs, AI operations
- The operational discipline for language-model systems — evaluation in CI, tracing, cost control and versioning — because these systems degrade quietly rather than failing loudly.
N
- NoOpsno-ops, fully managed platform
- An arrangement where developers ship code and someone else owns everything between the commit and the customer — including the 24x7 pager.
R
- Retrieval-Augmented GenerationRAG, retrieval augmented generation
- Fetching relevant information from your own data at question time and giving it to the model, instead of relying on what the model already knows.
S
- Site Reliability EngineeringSRE, reliability engineering
- Treating reliability as an engineering problem with measurable targets and a budget for failure, rather than as an aspiration to never break.
V
- Vector Databasevector store, embedding database
- A store for embeddings that finds text by meaning rather than by keyword — and for most workloads, a Postgres extension rather than a separate product.
A term we should add?
If a vendor used a word at you and you are not sure it meant anything, tell us and we will define it properly.
Or email [email protected] · we reply within 1 business day