AI Prompts for Workout Plans, Meal Plans and Health Admin

A plan starts from a measured baseline and the days that genuinely exist in the week. It progresses in small planned steps with an easy week every fourth, because most injuries in new training come from doing too much too soon.

• Screening before planning: chest symptoms, dizziness, injuries, conditions, pregnancy and medication are asked about first, and anything flagged holds the plan until a doctor has been spoken to.
• Three days, not five: she has two children under ten and a desk job. A plan built on five days would have been abandoned by week three.
• The goal is a distance, not a weight: no calorie target, no goal weight, no rate of loss. The measure is 10K without stopping, on 16 May.

A session is built backwards from the minutes available, and around what hurts rather than through it. Anything sharp, new, swelling or worsening ends the session and goes to a clinician instead.

• Dull and familiar, or sharp and new: the prompt asks which, because the answer decides whether there is a session at all.
• Effort in words, not zones: able to talk, able to say a sentence, unable to speak. No heart-rate straps required.
• The 15-minute version: written in advance, so a disrupted evening produces a short session rather than none.

A test is only useful if it can be repeated the same way: same route, same rest, same shoes, same time of day. Measurements are of what the body can do, and the tracker has no weight, body-fat or calorie column anywhere.

• Conditions recorded with the result: the park loop, 7am, a rest day before. Without them the second test measures the weather.
• Beyond normal variation: the prompt says which changes are real and which are within day-to-day noise, so the plan is not rewritten on a bad morning.
• The plan follows the numbers: strength was progressing, running was not, so the next block raised running volume and held the strength work.

A meal plan fails when it assumes an hour every evening, ignores what the household will eat, or bans the things people like. This one sets no calorie or macro targets, no weight-loss rate and no goal weight, and calls no food forbidden.

• Matched to the evening: Monday and Wednesday are the short evenings, so those are the 15-minute meals. That single change is what made the week survive.
• One dish, two finishes: where adults and children eat differently, the plan gives one dish finished two ways rather than two dinners.
• A hard limit in the prompt: if the request is for a very restrictive plan, or anything suggests disordered eating, the prompt stops and points to a doctor or a registered dietitian.

A list works when it is ordered the way the shop is laid out, quantities match the servings, and the cupboard has been checked first. Substitutions are decided in advance, because deciding in the aisle is how a plan becomes three unrelated purchases.

• Quantities per serving: stated so they can be checked, rather than guessed at the shelf.
• Swaps in advance: three items her shop often lacks came with a decided alternative, so a gap did not end the plan.
• A top-up list: the fresh things are bought midweek, which is why Thursday’s dinner still works.

Most households have more usable food than they think, and throw away what they could not see a use for. The prompt builds around what is there and what needs using first, and asks for nothing that cannot be substituted.

• What to use first: the spinach decided the dish, which is the opposite of choosing a recipe and shopping for it.
• Allergies restated: every allergy given is restated and checked against the recipe, with a note to check labels yourself.
• Food safety where it matters: rice, poultry and reheating carry specific notes rather than being left to assumption.

A preparation sheet states the problem in a sentence, gives the history in dates, and puts the most important question first. The prompt prepares facts and questions only: it does not diagnose, suggest a cause or recommend a treatment.

• The worry, said out loud: what she was actually afraid of went into the first minute, in one sentence, instead of the corridor afterwards.
• Questions that cannot be answered yes or no: including what happens if nothing is done, and what would change the advice.
• The after note: what was decided, who does it and by when, recorded the same day while it is still accurate.

A reference range is what a laboratory considers usual for most people, not a target, and it differs between laboratories. The prompt explains what each thing measures and shows the trend. It never says what a result means for your health.

• Described, not judged: a value is reported as in range, above or below. It is never called good, bad, worrying or fine.
• No treatment advice, ever: the prompt will not suggest starting, stopping or changing a medication or supplement. That belongs to the clinician who ordered the test.
• Far outside the range: the only thing the prompt will say is that it should be raised with a clinician promptly.

Routine checks are missed because nobody holds the list: dental, eyes, blood pressure, vaccinations and the national screening programmes all run on different intervals. Those rules differ by country and change, so everything is marked to confirm.

• Never told what you need: the prompt presents what is generally offered and sends the eligibility question to a clinician or the national programme.
• Spread on purpose: the booking calendar avoids six appointments in one month, which is how a schedule survives contact with a working year.
• Questions for the doctor: particularly where family history or a condition changes which checks apply, and from what age.

A large share of medical bills contain errors, and a denial is not a final answer. Most have a stated reason, a deadline and an appeal route, and a factual appeal with the evidence attached succeeds far more often than a phone call.

• The code meant something: CO-15 means the referral was missing or invalid. Hers existed but had not been included on the submitted claim, which is a fixable clerical failure.
• The duplicate nobody spotted: line 4 repeated line 2: same code, same date, same $210. It was only visible on the itemised bill, which had to be requested.
• Only from the documents: the prompt never guesses a billing code or a policy term, and marks unclear wording [CHECK POLICY].