Knowledge base

The terms behind working with AI.

Short, honest explanations of the terms you meet once AI starts doing real work. Each term in one paragraph, with a dossier next to it for those who want to go deeper.

The terms.

AI agent

A system that is given a task and chooses its own steps to complete it, within boundaries you set.

Read the explanation

Context engineering

Managing what an AI model sees at any moment: the skill that succeeded writing prompts.

Read the explanation

Harness

The machinery around an AI model that determines whether it delivers reliable work.

Read the explanation

Human-in-the-loop

The design principle where AI does the executing work and a human keeps the decision moments.

Read the explanation

Evals & guardrails

Measuring whether AI output is right, and guarding the boundaries within which a system may act.

Read the explanation

Prompt engineering

Formulating a good instruction for an AI model, and why that is no longer the whole story.

Read the explanation

Deeper into the technique? We wrote the five-part dossier series "Working with AI" on how agents really work. To the insights

Rather an explanation for the whole organisation?

A keynote or training makes these terms tangible in the language of your work.

Book a conversation