The four levels

The four levels of the course, Beginner, Intermediate, Advanced and Expert: what you can do at each step, the success criterion and the associated badge.

By QDNA · Published on September 9, 2026 · Updated on September 10, 2026

In brief. Each level describes what you will be able to do at the end of its lessons, and is measured on three skills followed throughout the course: frame (say clearly what you want), check (control an answer before using it) and steer (follow up, chain, equip your team).

A level is validated when, for each of its lessons, the quiz is passed and the proof of mastery is submitted. A badge is then issued. You choose the order of lessons within a level freely.

Level 1

Beginner

“I say clearly what I want”

Illustration of the Beginner level: a person writes a first complete prompt in front of Copilot Chat. — Illustration generated by AI

What the level contains: 6 lessons · 90 min of reading in total
What you can do: State a goal in one sentence with a precise action verb. Say for whom and for what. Specify the expected format and length. Paste or attach the text to process instead of describing it. Proofread the answer before using it.
Success criterion: Out of ten prompts in the portfolio, at least eight contain the four ingredients, and at least seven produce a usable output after at most one follow-up.

Level 2

Intermediate

“I give the model what it needs to do well”

Illustration of the Intermediate level: attached files, examples and rules given to the model to do well. — Illustration generated by AI

What the level contains: 7 lessons · 115 min of reading in total
What you can do: Attach the right files and say what to do with them. Provide one or two examples of the expected result. State positive constraints and "if... then" rules. Put the source at the end of the prompt. Iterate by changing one thing at a time. Tell the model what to do when information is missing.
Success criterion: On the same case, two versions of the prompt are produced and the difference in output is documented; at least three techniques of the level are visible in the final version; the example format is respected on three tries.

Level 3

Advanced

“I make it reason and I make it verify”

Illustration of the Advanced level: a chain of prompts, verified quotations and a review grid. — Illustration generated by AI

What the level contains: 6 lessons · 115 min of reading in total
What you can do: Break a complex task into steps or chained prompts. Ask for verbatim quotations before analysis. Require a "what I did not find" section. Analyse a data file with an explicit calculation plan. Write a review grid and apply it. Produce an iterated deliverable in Pages.
Success criterion: On a case of three documents, the synthesis links every statement to a verifiable quotation; a peer checks five statements at random and finds none false; the "not found" section is present and accurate.

Level 4

Expert

“I design prompts that others reuse”

Illustration of the Expert level: a shared prompt library, an agent and a test set. — Illustration generated by AI

What the level contains: 6 lessons · 120 min of reading in total
What you can do: Maintain a library of versioned prompts. Write meta-prompts. Build a simple agent with instructions, sources and examples. Run an A/B test on a small set of cases. Document the risks and train peers. Know when not to use Copilot.
Success criterion: A prompt or an agent is adopted by at least two colleagues; the associated test set scores at least 8 out of 10; the library sheet has a version, a date, a use case, a sample output, known limits and an owner.