Part one
AI workflows for real work
Choose a task that is already on your list. Each recipe takes you from source material to a checked draft you can actually use. Several assume you already prompt and need the result to survive the activity file.
The workflow in every recipe
- What you're making. The actual thing, and when this is the wrong approach.
- What to have open. The material you need in front of you before you start.
- The prompt. Copyable, with the bits to swap out marked in brackets.
- What good looks like. So you can tell whether yours is any good.
- How to check it. The specific step, not "use your judgment".
- Where it goes wrong. The particular way this job tends to fail.
- From needs assessment to learning objectives 25 min · Handle with care You have survey results, a chart review finding, and a guideline update. You need objectives that tie back to the gap and survive review.
- Turning 400 evaluation comments into something you can act on 30 min · High risk The open text box is where the real feedback lives and where it goes to die. This gets you themes, counts, and the quotes that matter, without quietly softening the criticism.
- The post-activity outcomes report 45 min · Handle with care Turn source material into a structured outcomes-report draft while keeping every figure and claim traceable.
- Faculty email you stop dreading 15 min · Low risk Invitations, chasing disclosures, deadline reminders, and the third follow-up to the busy cardiologist. The unglamorous volume work where your week actually goes.
- A facilitator guide from a deck somebody else built 35 min · Low risk You have 60 slides and a room in three days. Turn it into a guide with timings, discussion prompts, and a plan for when the session stalls.
- Rehearsing the conversation you have been putting off 20 min · High risk Use a fictionalized scenario to rehearse a difficult conversation, then review what you said and what you avoided.
- The weekly RSS packet 20 min · Handle with care Grand rounds is not one activity. It is a new session every week: announcement, disclosure reminder, session objective, and eval questions, without inventing a gap you do not have.
- Chasing disclosures without pasting the forms 25 min · High risk The volume work of Standard 3: reminders, a relevance screen from facts you already hold, and a mitigation note. The forms themselves stay out of the tool.
- Screen a talk for commercial bias 25 min · Handle with care You have an outline or a transcript, not a marketing deck. Flag promotional language, missing balance, product names, and logos before the session is accredited.
- Cases and items from a gap you already have 30 min · High risk Draft vignettes, questions, distractors, and rationales from the practice gap and the medicine you supply. The tool does not invent the clinical facts.
- An IME needs narrative from evidence you already hold 40 min · High risk Draft the gap, objectives, and outcomes plan for a grant or joint-providership proposal without taking the supporter’s language into the education.
- Program-analysis prose from counts you already have 40 min · Handle with care Turn PARS numbers, eval summaries, and last year’s notes into self-study or year-end narrative without inventing a trend or bumping an outcome level.
- PARS annual-close checking notes 35 min · High risk Turn a PARS or JA-PARS export into a close checklist and Program Summary prose without inventing a learner count or closing a series in the wrong year.
- Joint-providership walk-through 30 min · High risk Who owns the file, what each side may draft, and a planning memo that does not invent the accreditation statement or treat the arrangement as a legal partnership.
- Enduring-module outline from a gap you already have 35 min · High risk Outline an on-demand module with release, review, and termination dates, credit as time to complete, and learners you can verify. The tool does not invent the medicine or the PARS record.
- Supporter outcomes: promised level versus measured 40 min · High risk A 90-day outcomes dialect for grant-funded education: the Moore level you promised, the level you measured, and a number audit that will not promote competence into performance.
- Does this count? A process, not a yes 25 min · High risk Build a trail of where you looked for credit, MOC/CC, and state-board questions. The tool helps you log the search. It does not decide whether the activity counts.