Glossary

Software that can reason about a goal, decide what steps to take, use tools to act in the real world, and adjust based on the results — with minimal human hand-holding.

The use of AI — especially machine learning and AI agents — to detect, diagnose, and resolve operational problems with far less human intervention.

A pipeline that automatically tests, builds, and deploys code on every push, removing the manual steps between a Git push and a live release.

The practice of shipping software reliably and quickly by automating everything between "code is written" and "code is running in production."

Building security into every stage of a CI/CD pipeline — dependency scanning, secrets management, and automated checks — instead of bolting it on at the end.

A tool for packaging an application and its dependencies into a container, so it runs the same way in every environment.

Describing cloud infrastructure in code (typically with Terraform) instead of clicking through a console, so environments are repeatable, reviewable, and recoverable.

The standard system for orchestrating containers at scale — handling deployment, scaling, and recovery of pods and services.

An open standard that lets AI agents connect to external tools and data sources through one consistent interface, instead of a custom integration per tool per agent.

The practices for deploying, monitoring, and maintaining machine learning models in production — logging what a model does and handling its failures gracefully.

The combination of logging, metrics, and monitoring that lets an engineer know when something breaks — ideally before users do.

Building internal developer platforms that give every engineer in an org a self-service "paved road," rather than each team automating its own deployments separately.

A pattern where an AI system retrieves relevant documents and hands them to a model as context, instead of relying on what the model already "knows."

An engineering workflow where you write a clear, reviewable specification of what to build and why before any code is generated, then build against that spec as the single source of truth.

Prompting an AI in loose natural language, accepting whatever it generates, and re-prompting until it seems to work — with no written specification guiding the process.