Prompt, Context, Loop and Graph Engineering

Prompt engineering, context engineering, loop engineering, graph engineering. What do these mean, and how do they connect?

Let's take one task, a weekly sales report, and see where each one fits.

What is prompt engineering?

A prompt is the instruction you give AI, and prompt engineering is writing it well. "Write a sales report" gets you something generic. A better prompt says what you want, who it is for and what good looks like, like "a one-page report on the top three regions, in simple words for senior leaders."

What is context engineering?

Context is everything AI sees, including your prompt, files, past chat and memory. Context engineering is choosing the right information to include. For the report, AI needs this week's numbers and last week's report, not the whole year's data. With too little, AI guesses, and with too much, the important parts get lost.

What is loop engineering?

A loop is a cycle of doing, checking and fixing. AI writes the report, then checks whether the totals match the sales file and whether it fits on one page. If not, it fixes the draft and checks again, until every check passes or it reaches a set number of tries.

What is graph engineering?

A graph is a map of all the steps in a process and where the work goes next. AI drafts the report, then reviews it. If the review finds problems, it goes back for changes, and if not, it moves to formatting. Each step can run its own loop.

How do these four connect?

Each one sits inside the next. Every step in the graph runs a loop, every try in the loop uses context, and every context starts with a clear prompt.

What to remember

A clear prompt tells AI what you want, context gives it the right information, a loop checks the work and a graph organises the whole process. New buzzwords will keep coming, but if you understand these four ideas, you will know how to get good work from AI.