Start with the job to be done

The working assignment is to turn a verified meal-prep class brief into a truthful two-variant email test for busy beginners considering a local class. That sentence is narrower than “use AI for marketing.” It identifies a deliverable and a reviewer. Write a definition of done with three layers: the output must satisfy the audience's need; factual or functional claims must be traceable; and a named person must own the final decision. Generation volume is not strategy. A useful campaign has a defensible promise, a specific audience action, and a test that can teach one thing.

Start by separating tasks. The model may draft, classify, transform, compare, or suggest. It may not silently approve, publish, grade, deploy, cite, or consent on someone's behalf. For this assignment the authoritative material is approved product facts, audience problem, channel, word limit, brand voice, prohibited claims, CTA, and test hypothesis. Anything absent from those inputs is either an explicit assumption or an unanswered question.

Use this prompt as a realistic starting contract:

Use only these facts: two-hour in-person class; ingredients included; participants leave with five recipes. Audience: busy cooking beginners. Draft a subject line and email under 140 words with one sign-up CTA. Do not claim savings, health outcomes, superiority, scarcity, or testimonials. Then make variant B by changing only the angle from convenience to confidence.

Notice what the prompt does: it states the setting, limits the output, names forbidden behavior, and requests evidence that can be reviewed. It does not ask the model to “make it amazing.” If a constraint matters, make it testable. Replace “be accurate” with a source boundary, formula check, test command, rights ledger, or approval step.

A useful response would look like this: Two compact emails with the same offer and CTA; each claim maps to the brief, while the changed angle is explicitly labeled as the sole test variable. That description is intentionally observable. “Looks good” is not acceptance. The operator must trace every factual phrase to the approved brief, check link and disclosure requirements, review accessibility and brand voice, then measure the predeclared conversion event. Keep the source material beside the draft so review means comparison, not memory.

Run the work through brief → claims → draft → review → test → learn. Stop when an authoritative input is missing, a high-risk claim lacks evidence, private material cannot be safely removed, or the proposed action exceeds the permission granted. Escalation is successful workflow behavior, not model failure.

Common framing mistakes are fabricated urgency; fake statistics; multiple variables changed; vague CTA; stereotyped audience language; optimization without a hypothesis. Prevent them by writing a one-paragraph job card: user, decision, deliverable, source of truth, constraints, reviewer, and stop condition. This card becomes the anchor for every later prompt.