How We Optimize AI to Manage Our Rare Life Remix
From Medicaid maps to insurance appeals, appointment follow-up and planning for the future, we use AI to make complicated work more manageable—without handing over human decisions.
Rare disease has a visible life and an invisible one
People see our daughter’s appointments, equipment, medications and milestones. They do not always see the second shift that begins after bedtime: reading a denial letter line by line, hunting for the missing form, comparing waiver language, remembering which call has to happen first, and trying to explain the whole thing to the person you love when both of you are already tired.
That invisible work can make someone feel like they’re carrying a hundred open tabs in their head. Some are urgent. Some are emotional. Some are just small enough to be forgotten until they become expensive, late, or impossible to ignore.
AI has not made that work disappear. It has helped us hold it with a little less friction—and, on good days, a little more breathing room.
The first map that made us exhale
One of the first genuinely useful things I asked AI to make was a map of Indiana Medicaid and the Health & Wellness Waiver. We knew many of the nouns: FSSA, Medicaid, BDS, case management, home health, Structured Family Caregiving. What was harder was keeping the relationships straight, especially when each phone call seemed to add another layer.
So I gathered the official guidance, added our own notes, and asked AI to turn the pieces into a picture. The result did not change our daughter’s eligibility, make a provider call us back, or guarantee a service. It gave us, as a couple, a shared view. We could point to one part of the page and say, “That question belongs with the case manager,” or, “This is on the medical side, not the waiver side.” We stopped having to rebuild the whole system from memory every time we talked about it.
Figure 1. A simplified family-facing map of Indiana Medicaid and the major BDS waiver lanes, with our daughter’s H&W lane highlighted.
The public version of a map does not need every identifying detail. Our working version can be specific; the version we share uses first names and generic descriptions such as “waiver case manager,” “Structured Family Caregiving provider agency” and “home health or nursing agency.” That protects privacy and also makes the picture more useful to other families, whose provider names will be different.
Figure 2. One family’s support lane, using first names and generic provider categories rather than company names.
What I am actually asking AI to do
At its best, AI plays four modest roles for us. It translates language that was written for agencies or insurers. It organizes scattered information into a shape we can use. It drafts a starting point when a blank page feels impossible. And it helps us remember what happened, what is still open, and who owns the next step.
It is not the doctor, case manager, attorney, financial adviser, or parent. It does not know our daughter the way we do, and a polished answer is not the same thing as a correct one. I get the most value when I give AI a narrow job, good source material, and an explicit instruction not to guess.
A PROMPT I USE
Using only the documents I provide, create a plain-English map of this program. Show who makes each decision, where the family enters the process, and what still needs to be verified. Cite the source for every important rule. Do not fill gaps with guesses.
Make responsibility visible without losing the person
A chart can become just another piece of bureaucracy if the child disappears inside it. I like to create a second view that asks a more human question: who is responsible for what, and how does each role support life at the center?
That view helps in practical ways. It shows who can write an order, who can authorize a waiver service, who coordinates the plan, who delivers the support, and who carries the daily knowledge that no assessment can fully capture.
Figure 3. A “who does what” map that keeps the child—not the paperwork—at the center.
A glossary is the other companion piece. Acronyms create unnecessary distance. Once every acronym has a plain-English job description, we can use the same language in calls, emails and meetings without feeling as though we are taking a test.
Figure 4. A plain-English glossary for the Indiana waiver terms that appear most often in our family’s conversations.
Turn an insurance denial into a path forward
A denial letter can make a family feel as though the answer is simply no. Usually, several smaller questions are hiding inside that “no.” Is the plan calling the request a benefit exclusion? Is it disputing medical necessity? Was a prior authorization missing? Is the problem coding, network status, or documentation?
AI is useful here because it can slow the letter down. I can ask it to extract the reason, deadline, policy language, missing evidence, and next actions. Then I can build an evidence map: what the insurer requires, what we already have, what is missing, and who can supply it.
Figure 5. A practical insurance-appeal path, from the denial notice to the next available review right.
A PROMPT I USE
Read this denial as an organizer, not as a lawyer or clinician. Extract the reason, the deadline, the policy criteria, the evidence already in the record, the missing evidence and the next actions. Use only the materials I provide, and flag anything that needs human confirmation.
Our family still has to verify every date, quotation, diagnosis, and factual statement. I never want AI to make an appeal sound stronger by inventing a symptom, treatment, or conversation. A clear appeal built from real facts is far more useful than a dramatic one built on shaky ground.
That principle applies well beyond insurance appeals. Sometimes an approval or denial comes down, at least in part, to how clearly we describe our daughter’s needs.
We have wrestled with this problem many times. Lindsey wants our daughter to be seen as the extremely smart, funny, and sassy girl she is. To be clear, I do too. We never want a diagnosis or a list of limitations to become her entire story.
But when a state assessor is sitting in your home to determine whether your child qualifies for a waiver—especially after more than two years on a waitlist and countless appeals—the words you use matter. Your examples matter. The details you leave out matter.
As parents, we all naturally focus on progress. We celebrate what our children can do. We adapt so frequently that extraordinary levels of care can begin to feel ordinary. We may say, “She tells us what she wants,” without explaining that only people who know her well can interpret her eye gaze, expressions, sounds, and routines. We may say, “She helps get herself dressed,” without explaining that she still requires hands-on assistance, physical positioning, extra time, and close supervision throughout the process.
Both statements may be true. But they do not communicate the same picture to someone deciding whether support is medically or functionally necessary.
For example, an assessor might ask:
“Can your daughter communicate her needs?”
My first instinct might be to say:
“Absolutely. She is incredibly smart and expressive. We usually know exactly what she wants.”
That answer reflects who she is, but it does not fully describe the support she needs. A more complete answer would be:
“She understands far more than she can reliably express. At home, we often recognize her eye gaze, facial expressions, vocalizations, and familiar routines because we know her so well. Someone who does not know her may not understand those signals. She cannot consistently communicate pain, illness, danger, hunger, or an urgent need without an attentive and trained caregiver.”
The second answer does not diminish her intelligence. It does not exaggerate her disability. It simply gives the assessor the functional information needed to understand her safety and support needs.
This is one place where I use AI as a rehearsal partner. I may ask it to pose the kinds of questions a waiver assessor or caseworker is likely to ask. After I answer in my normal language, I ask AI to point out where I have been vague, where I may have minimized the amount of assistance involved, or described an adapted success without explaining the support that makes it possible.
The goal is never to manipulate the process or make our circumstances sound worse than they are. The goal is to tell the whole truth in language the system can understand—while still honoring the whole person our daughter is.
Help the clinician see the whole person
A Letter of Medical Necessity belongs to the treating clinician. The clinician must agree with it, edit it, sign it, and own the medical judgment. But families often carry the scattered pieces the letter needs: the denial reason, therapy evaluations, device specifications, previous treatments, daily safety concerns, and the small details that show why a standard alternative is not actually an alternative.
AI can arrange those pieces into a clinician-editable outline. The job is not to make our daughter’s life sound worse. It is to make sure the reviewer sees the parts of her life that do not fit neatly inside a diagnosis code.
Figure 6. The six connections a strong Letter of Medical Necessity usually needs to make.
A PROMPT I USE
Create a clinician-editable outline using only these records. Put brackets around anything the clinician must confirm. Do not invent symptoms, measurements, treatment history, study citations or medical conclusions.
Hold onto what happens between appointments
Appointments compress a lot of life into a very small window. We arrive with months of observations, try to explain what has changed, make decisions in real time, and leave with orders, referrals, and follow-up tasks that may depend on three different offices.
Before the visit, AI can help me make a one-page brief with the top three questions, recent changes, and records we need. During the visit, I capture decisions, owners, and dates. Afterward, I turn the notes into a follow-up list: who is requesting the prescription, when the lab should be repeated, which appointment needs to be scheduled, and what to do if the order stalls.
Figure 7. A before-during-after-follow-through loop for medical appointments.
The final reminders go into a real calendar or task system. I do not let an important deadline live only inside a chat. AI helps extract and organize the work; our shared calendar is where the family can trust that it will reappear.
Prepare for the future without pretending we can predict it
Estate planning is technical, but it is also deeply personal. The questions may sound financial or legal, yet underneath them are questions about who knows how our daughter communicates, what makes her feel safe, who could step in, where she might live, and how the people around her will preserve a life that feels like hers.
AI should not choose a trust, write documents for us to sign or replace a special-needs planning attorney. It can help us arrive at the meeting with a thoughtful inventory: current accounts and beneficiary designations, possible decision-makers and backups, future care assumptions, benefits questions, housing ideas, and the beginnings of a letter of intent.
Figure 8. An estate-planning conversation map that includes the human details alongside the legal and financial ones.
The most important part of that preparation may not be the asset list. It may be the practical, loving knowledge future caregivers would otherwise have to learn in a crisis.
Make each resource go farther
“Financial efficiency” can sound cold when the subject is caregiving. For us, it means something much more human: not losing money to a missed reimbursement, not choosing an insurance plan without understanding the real annual exposure, not forgetting a renewal date, and not spending hours recreating a calculation we already made six months ago.
AI can compare scenarios, expose assumptions, and build a renewal calendar. It can place premiums, deductibles, out-of-pocket limits, employer benefits, waiver supports, grants, travel costs, and caregiver time in the same picture. I ask it to separate known costs from estimates and to show the calculation behind every total.
Figure 9. A family-centered resource map for comparing costs, timing, deadlines, risk and flexibility.
The output is a decision aid, not a recommendation. Benefits rules, tax treatment and investment questions still belong with the qualified professionals who can see the full picture and take responsibility for the advice.
The guardrails are what keep this human
Below is a list of helpful tips for when someone is using AI:
Keep the source in view. The original denial, medical record, policy, agency notice or legal document outranks the AI summary. I ask for page references and direct quotations only when the source is in front of the model.
Date the work. Benefits programs, provider arrangements and family circumstances change. Every map and checklist should say when it was made and what needs to be rechecked.
Protect privacy. Before I upload anything, I remove names, dates of birth, addresses, member numbers and other identifiers the task does not require. I also think about the privacy settings and terms of the tool I am using.
Let professionals own professional judgment. Clinicians own medical conclusions. Attorneys own legal advice. Financial and tax professionals own recommendations in their fields. AI can help us prepare, but it does not get the final word.
Read it like a parent, not a proofreader. I ask whether the output sounds like our actual life. Is anything exaggerated? Is something important missing because it was never written down? Does the plan leave room for our daughter’s dignity, preferences, and joy?
A simple place to start
A family does not need to build all of this at once. Start with one problem: a confusing denial, an upcoming appointment, a benefit renewal or a stack of documents that do not seem to fit together. Remove unnecessary identifying information. Give AI a narrow job. Ask it to separate facts, questions and next actions. Then read the result with another human who knows the situation.
A REUSABLE STARTING PROMPT
Help me organize this situation, not decide it for me. Use only the materials I provide. Separate confirmed facts, reasonable inferences and questions that require verification. Identify deadlines, missing information, decisions and next actions. Do not invent facts, rules, medical findings or dates.
Turn every next action into a person and a date. Save the final version somewhere both adults can find it. The real value is not producing one perfect document; it is reducing the number of times the family has to start from zero.
More room for the parts that matter
The best use of AI is not the cleverest prompt or the most elaborate chart. Sometimes the result is a calmer email. Sometimes it is remembering the one question we did not want to lose before an appointment. Sometimes it is ten minutes at the kitchen table where we are both looking at the same page instead of carrying different pieces in our heads.
Those may sound like small wins, but a “remixed life” is built from small wins. Every family’s remix will require different elements. The point is not to copy our exact setup. It is to find one piece of heavy work that could be made lighter.
We are not trying to automate our daughter’s life or outsource our instincts. We are trying to spend less of it translating systems—and more of it being her family.
Notes for Publication
Reader note
This article describes one family’s workflow. AI outputs can be incomplete or wrong, and medical, insurance, Medicaid, waiver, legal, tax and financial rules change. Nothing here is medical, legal, financial or benefits advice.
Indiana source note
The Indiana diagrams are simplified, family-facing interpretations based on official materials available in June 2026: Indiana FSSA — Medicaid HCBS Waivers; Indiana FSSA — Medicaid HCBS Programs; Indiana FSSA — BDS Waiver Reset; and Medicaid.gov — Indiana Waiver Factsheet. Indiana is redesigning parts of its waiver system, so terminology and administration should be verified.
RELATED ARTICLES
