Competency guide
Getting real work done
From impressive draft to finished, owned, professional output.
AI produces drafts quickly. Professionals produce finished work: verified, tuned to the audience, and owned completely by the person who ships it. The delivering competency is the bridge — knowing what the last 20% is, why it belongs to you, and how to spend AI acceleration where it actually compounds.
- Stage 1 · E1 FOUNDATIONAL
Own everything you ship
The accountability rule is absolute: work you ship is yours, however much of it a model drafted. Review AI-assisted output as if a talented but unaccountable assistant wrote it — because that is exactly what happened.
Learn where the last 20% lives: it is usually the judgment, the stakes, and the audience nuance. The model gets the structure and the bulk right; the parts that decide whether the work lands are the parts you finish by hand.
Try it this week
On your next AI-assisted deliverable, before sending, list what you changed from the draft and why. If the answer is "nothing," read it once more as a skeptic — there is almost always something.
- Stage 2 · E2 PROFICIENT
Spend acceleration where judgment lives
With two hours for a day-sized report, the strong move is not "AI writes it, I fix typos." It is: AI structures the outline and drafts sections from your source material, and you spend the saved time reworking the parts where your judgment adds value — the recommendation, the framing, the parts a reader will challenge.
Keep the structure, verify the evidence, remove unsupported claims, tune the story to the decision the audience faces. That editing pass is the job now.
Try it this week
Time-box a real deliverable: 30% of the time for AI drafting, 70% for your verification and rework. Compare the result to your usual split.
- Stage 3 · E3 DISTINGUISHED
Pick the right things to accelerate
Not every task rewards AI equally. The strongest candidates are high-volume, pattern-shaped, verifiable work — the weekly report with fixed structure whose output you can check quickly against source data. The weakest are one-off judgment calls resting on unwritten context.
Distinguished operators also use AI to raise quality, not just speed: generating the counter-arguments before the meeting, stress-testing the plan for edge cases, rehearsing the skeptical reader. Same deadline, stronger work.
Try it this week
List your five most frequent deliverables. Rank them by (volume × verifiability) and pick the top one as your next AI-first candidate — and notice which one you should keep fully manual, and why.
- Stage 4 · E4 EXCEPTIONAL
Deliver work that teaches the standard
At the top of this competency, your AI-assisted work is indistinguishable in accountability from your solo work and better in reach: more alternatives considered, more objections pre-answered, more evidence traced. Colleagues cannot tell where the model ended and you began, because you never shipped anything you had not made yours.
You also model the honesty: where disclosure of AI use is required or expected, you disclose without drama. Trust compounds; concealment risks the relationship and sometimes the contract.
Try it this week
Take a finished piece of AI-assisted work and write the two-sentence provenance you would give if asked directly how it was made. If writing it makes you uncomfortable, the work is not done.
On the exam
Delivery scenarios grade output quality, described process, and ownership — pasted-verbatim answers with no evidence of review score low even when the text reads well.
Ready to see where you stand? The free check scores all six competencies in about fifteen minutes.