This isn't harder prompts. It's the shift from using AI well to building repeatable tools with it — and developing the judgment to know exactly where it shouldn't be making the call.
This level assumes Practitioner's integrated workflows are already routine. The work here is turning your own habits into something reusable — and building the judgment to know when not to use them.
Gather every prompt you've reused more than twice over the last month into one doc — mess and all. Lesson 1 works directly from this.
Pick one upcoming planning or budget conversation to use as your working example in Lesson 2.
Think of one recent moment you weren't sure whether to trust AI's suggestion or your own judgment. Lesson 4 comes back to this directly.
There's no single "right" system here. The goal by the end of this module is that you leave with your own way of working — not a copy of someone else's.
By this point you've got a dozen prompts you reuse in slightly different forms every week — copy-pasted from old chats, never quite organized.
Turn your actual working prompts into a personal, versioned template library — with your brand voice, your reporting cadence, and your team's terminology built in, so every new prompt doesn't start from zero.
Here are five prompts I use regularly, pasted below. Rewrite them as a reusable template set: replace the specific details with clearly marked placeholders, and add a one-line note under each explaining when to use it.
Building the perfect library once and never updating it. A template library that isn't revised as your role or tools change becomes dead weight within a quarter.
"What happened" reporting — Foundation, Practitioner — looks backward. Budget planning and target-setting requires looking forward, under real uncertainty.
Use AI to build out budget scenarios and sensitivity checks — "what if CPMs rise 15%" — as a real input into planning conversations, not just historical analysis dressed up as a forecast.
Our current monthly budget is [X] at an average CPM of [Y]. Model three scenarios: CPMs rise 15%, CPMs fall 10%, and budget increases 20%. For each, estimate the impact on volume and give one recommendation for how we'd adjust strategy.
Presenting AI-generated forecasts as precise predictions rather than what they are — directional scenarios to stress-test a plan, not numbers to commit to publicly.
Everyone on a team ends up with wildly different personal habits — some great, some sloppy — and quality becomes inconsistent depending on who's doing the work.
A lightweight way to document and share your own working methods — from Lesson 1's library — so a team converges on a consistent standard, without turning it into unused bureaucracy nobody opens twice.
Turn this prompt library [paste from Lesson 1] into a one-page onboarding guide for a new team member: what each template is for, one example of good output, and one common mistake to avoid when using it.
Mandating a rigid process nobody actually follows, instead of sharing a living, editable resource people will genuinely use and improve.
The more fluent you get, the easier it is to over-trust AI on calls that actually need human judgment — a final budget sign-off, a sensitive client relationship decision, a brand-risk call.
A sharper, more strategic version of Foundation's "what not to hand over" — this time about judgment calls, not just data privacy. Use AI to lay out considerations; keep the actual decision with a person.
List the factors for and against [a real decision you're weighing — a budget cut, a client relationship call, a brand-risk judgment]. Don't recommend a choice — this decision is mine to make, I just want the considerations laid out clearly.
Treating fluency with AI as a substitute for accountability. The more comfortable you get, the easier it is to let a confident-sounding suggestion quietly become the decision, instead of staying an input to it.
New AI features and tools launch constantly, and chasing every one of them is its own time sink — a full-time job on top of your actual full-time job.
A practical filter for evaluating whether a new AI feature or tool is actually worth changing your workflow for, versus noise. This is the meta-skill that keeps everything else in this course from going stale.
Here's a new AI feature or tool I'm considering: [description]. Compare it against what I currently do for [specific task]. Would switching save meaningful time or improve quality, or is this a marginal difference dressed up as a big change?
Adopting every new tool release out of fear of falling behind. Most releases are incremental, and constant tool-switching costs more time than it saves.
The more fluent you get, the more invisible your reliance on AI becomes — even to you. Periodically ask yourself which of your "own" judgment calls were actually AI's suggestion, restated confidently. That awareness is the real skill at this level, more than any single workflow.
This is the last level — from here, the work is maintaining and sharing what you've built, not learning something new.
