AI Basics
A practical guide to the building blocks behind modern AI—what they mean, how they connect, and why they matter when moving from an idea to a reliable service.
AI that works through a task in repeated steps, rather than giving one immediate answer.
Repeat until the goal is reached
Useful for work that needs planning, checking, and improvement—such as research, support resolution, or report creation.
A common way for an AI assistant to connect with the tools and information it needs.
One connection standard can let an assistant work with email, documents, databases, and business software.
Several specialised AI agents working together, each responsible for a distinct part of the job.
Complex work can be divided between researchers, analysts, writers, and reviewers before one result is delivered.
A single control point between your applications and the different AI models they use.
Teams can switch providers, manage access, control usage, and monitor costs without rebuilding every application.
The cost of asking a trained AI model to process information and produce an answer.
Model size, response length, speed, hosting, and repeated requests all affect the cost of running AI at scale.
Repeatable tests that measure whether an AI system is accurate, useful, safe, and consistent.
Clear tests turn a promising demonstration into evidence that a system is ready for real customers and workflows.
Rules and checks placed around an AI system to reduce unsafe, unsuitable, or sensitive outputs.
Guardrails help protect customer data, brand standards, regulatory obligations, and the people using the system.
The records, measurements, and traces that show what an AI system did and why.
Teams can investigate failures, understand costs, measure quality, and improve a live AI service over time.