Steal My Prompt Vol. 54: The Open Enrollment Optimizer

The open enrollment email gives you two weeks. You log in, see two or three health plans with a price per paycheck next to each, and pick the one you had last year, because working out the rest would take an evening you do not have. The paycheck price is only part of what a plan costs you. The rest depends on the deductible, the out-of-pocket maximum, and how much care your household uses, which is the math most of us skip.

A fair comparison prices each plan for three versions of next year: a quiet year with only checkups, the year you expect, and a bad year where you hit the out-of-pocket maximum. The plan that costs least in a quiet year can cost more in a bad one, so the right choice depends on which year you are most likely to have.

That question is mine this month. Our baby is due in November, and the due date matters as much as the premiums. If the plan year starts in January, a November birth lands on this year’s plan and its deductible, so the most expensive part of having a baby is not what I am choosing for. The plan I pick now covers the baby’s first full year. A due date in early January would flip that and put the delivery on the new plan.

Give this prompt your plan numbers and the care you expect. It asks only for what is missing, prices every plan for all three versions of the year with the math shown, and puts each known change in the plan year where it lands. It ends with the one question that decides between the plans and leaves the choice to you.

What You Can Use This For

• Choosing between a PPO and a high-deductible plan with an HSA when the paycheck price is all you have compared
• A year with a change you can see coming, such as a baby, a planned surgery, or a new specialty prescription
• A chronic condition with regular specialist visits and refills, where hitting the out-of-pocket maximum is likely
• Deciding how much to put in an HSA or FSA once you know which plan fits

How to Use It

1. Open your AI assistant in an ordinary text chat.
2. Gather each plan’s numbers from your enrollment portal or its Summary of Benefits and Coverage, including the premium for your coverage tier, the deductible, the out-of-pocket maximum, coinsurance, copays, and any employer HSA or HRA money.
3. Describe the care you expect in plain terms, such as a specialist every other month or a prescription you refill monthly. If your insurer’s website shows last year’s claims, add what your care cost last year.
4. Copy the prompt, fill in the three bracketed fields, and send it. Answer its questions one at a time, or type skip to continue on labeled assumptions.
5. Check one row of the math with a calculator, then work through the verify list before you enroll.

Pro tip: Every Summary of Benefits and Coverage ends with the same coverage examples, including one for having a baby. They assume self-only coverage, so treat them as a side-by-side check of how each plan handles an identical delivery, not as your family’s cost.

The Prompt

I am choosing a health plan during open enrollment, and I want to see what each option would actually cost me next plan year before I decide. Do not choose the plan for me. Show me where each plan wins and what would change the answer.

Who will be covered, and any change I already know about for next plan year, with dates (a baby due, a planned procedure, a new prescription, someone leaving the plan): [WHO IS COVERED AND KNOWN CHANGES]

The care I expect next plan year, in plain terms (regular visits, therapy, prescriptions, planned procedures), what my care cost last year if I know it, and the doctors I want to keep: [EXPECTED CARE]

The plans I can choose from, using the numbers from my enrollment portal or each plan’s Summary of Benefits and Coverage (premium per paycheck or per month for my coverage tier, deductible, out-of-pocket maximum, coinsurance, copays, and any employer HSA or HRA contribution): [PASTE THE PLAN DETAILS]

Read everything before you respond. If something missing would change the comparison, such as my coverage tier, when my plan year starts, or what I pay in premiums and how often, ask me for it. Ask one question at a time, wait for my answer, and ask no more than four. If I type skip, continue with clearly labeled assumptions.

Every number you use must come from what I gave you or be labeled as your assumption. Never assume a plan rule I did not give you. Where a plan detail is unclear or missing, list it as something to confirm instead of filling it in. To build my expected year, apply each plan’s copays, deductible, and coinsurance to the care I described. Where that depends on the price of a service I did not give you, use a clearly labeled estimate and say how much the answer would move if the estimate is off.

Then give me five things, using tables and short bullets.

1. A cost table for each plan under three cases: a light year with only preventive care, my expected year, and a bad year where I reach the out-of-pocket maximum. For each case, show annual premiums, my out-of-pocket costs, any employer HSA or HRA money as a credit, and the total. Show the inputs behind every line so I can check the math. Leave tax effects out of the totals and say so.
2. The break-even point between the plans, in plain sentences, such as “Plan A costs less unless your care for the year passes about $X.” Say in plain words what that dollar figure measures, so I can judge whether my year is likely to pass it. If one plan costs less at every level of care, say that instead.
3. A timing check. Place each change I mentioned in the plan year it falls in, and say which plan year’s deductible and out-of-pocket maximum it counts toward. If a change lands in my current plan year rather than the one I am choosing, say so.
4. A list of things to verify before I enroll: whether my doctors and hospital are in network for each plan, how each plan covers my prescriptions, any plan rule you could not confirm from what I gave you, and the deadline and steps for adding a newborn or other new dependent, which I will confirm with HR.
5. HSA and FSA notes for each plan, written as rules to confirm with my benefits administrator. An HSA requires an HSA-eligible high-deductible plan. Being covered by a general-purpose health FSA, including one through my spouse, generally blocks HSA contributions, and a limited-purpose FSA does not. A dependent care FSA is separate and pays for a dependent’s care while I work. Do not state contribution limits unless I gave them to you.

End with which plan costs least in each of the three cases and the single question whose answer would decide between them for me. Do not tell me which plan to pick, and do not give tax advice beyond the rules above.

The podcast Explore AI Out Loud Find practical ways to use AI in your work and everyday life. Learn from real experiments, including the mistakes, so you can spend less time guessing and more time trying something useful. New episodes every other Tuesday.

Subscribe on YouTubeFollow on SpotifyFollow on Apple Podcasts

Latest episodes and the show

The podcast Follow Explore AI Out Loud Short clips from each episode, plus what did not make the cut. New episodes every other Tuesday.

Follow on InstagramFollow on TikTok

Full episodes and the show

Transparency and Notes

• Tested with Claude Opus 5.5 on made-up families and plans, including one with the premiums left out, which it asked for instead of guessing. Every total matched an independent recalculation. Other assistants have not been tested.
• Privacy. Leave out names, member ID numbers, and dates of birth. Check your assistant’s data settings before you describe health conditions.
• Educational only, not financial, tax, or medical advice. Your plan documents and benefits administrator decide what is covered and how long you have to add a new dependent.
• The pattern is Price the Bad Year. Compare options across a quiet, an expected, and a bad case, not the headline price. It works on a car insurance deductible, a software plan tier, or two job offers.
• Pairs with Vol. 43, The Insurance Denial Decoder, for the day coverage says no.