The basics
Three terms, one picture
Most of the current AI conversation fits in three ideas, stacked on top of each other.
-
AI — the model
Software trained on enormous amounts of text until it becomes very good at understanding and producing language. On its own it can only do one thing: read text in, write text out. It cannot open a file, check a calendar, or touch any system.
-
AI agent
A model that has been given a goal, plus software built around it so it can take real actions — search, read documents, update a record — step by step, instead of just answering a single question.
-
Agentic AI
The working pattern that makes agents useful: think about the next step, act, look at what happened, repeat — until the goal is reached. With humans deciding the limits: what it may do, and when it must hand over.
The topics
Pick a topic
Each topic is a short, self-contained explainer that ends with the design angle: what it means for the customer experience we own.
Topic 01
From prompts to harnesses
Prompt engineering, context engineering, harness engineering — the craft of working with AI keeps moving outward from the model. What each era means, and why every move looks more like design.
Read the explainer →Topic 02
What is context engineering?
An AI model only knows what it is shown, the moment it is shown it. Deciding what the service shows it — and what it must never see — is experience work.
Read the explainer →Topic 03
What is an AI harness?
The behaviour of an AI agent is decided mostly by the structure built around the model — its instructions, tools, memory and guardrails. Every one of them is an experience decision.
Read the explainer →Topic 04
What are evals?
How do you know an AI agent is any good — and who decides what “good” means? Why writing the quality bar for an agent is service design work.
Read the explainer →More topics will follow as the deep dive continues.