The method.

If you came from the Start here page, you already used this method without knowing it. This page names the seven steps behind those ready-made messages, so you can apply them to any health question. It takes five minutes to read, and it will change the quality of every answer you get.

Step 1

First, check whether this is a job for AI at all.

A few situations need a human being now, not a better answer. If you or someone near you has crushing chest pain, sudden trouble speaking or moving one side of the body, severe difficulty breathing, heavy bleeding that will not stop, a serious allergic reaction, or thoughts of ending your life, contact emergency services or get to any person who can help, immediately. AI can wait. These cannot.

AI is also the wrong tool when the answer requires hands or instruments: an examination, a test, a prescription. It can tell you what a test measures and whether your situation typically calls for one. It cannot run the test. Knowing the boundary is not a limitation. It is the first skill.

Step 2

Give the machine your context.

A generic question gets a generic answer. The single biggest upgrade you can make is to tell the AI who is asking: your age and sex, your known conditions, the medications and doses you take, what happened and when, what you have already tried, and what specifically worries you. You know things about your body that no one else knows. That knowledge is your contribution to the answer.

One caution. A chatbot is not your medical record, and what you type into a commercial AI is not protected the way hospital records are. Share what the question needs, and leave out what it does not: your name, your address, anything that identifies you.

Step 3

Ask the real question.

Ask the question you would ask at 2 a.m., in your own words, in your own language. Not the polished version. The real one. And ask for understanding, not just verdicts: "Explain this lab report line by line, in plain words." "What are the possible explanations, from most to least likely, and what makes each more or less likely in my case?" "What would make this more worrying? Less?" "Explain it again as if I had no medical knowledge at all."

You can ask the same thing five times, five ways, at three in the morning, without ever feeling rushed, judged, or billed. That is the one advantage you have over every clinic in the world.

Step 4

Question the answer.

Never accept the first answer. This is the step that separates using AI from being used by it. AI writes with total confidence whether it is right or wrong. Fluency is not accuracy. So push back, every time: "What is the evidence for that?" "What is the strongest argument against what you just told me?" "What are you uncertain about?" "What did you assume about me that I did not tell you?"

Good answers survive these questions. Weak answers fall apart, and watching an answer fall apart is not a failure. It is the system working. You just saved yourself from acting on something wrong.

Step 5

Get a second opinion from a second machine.

Doctors get second opinions. So should you, and yours are free. Take the same question, with the same context, to a different AI. Where the answers agree, you are probably looking at settled knowledge. Where they disagree, you have found exactly the spot that deserves more digging, or a human if you can reach one. Disagreement between machines is not noise. It is a map of where the uncertainty lives.

Step 6

Separate understanding from acting.

Here is the line that keeps you safe: learning carries almost no risk, acting carries real risk. You can read, ask, compare, and understand all night without danger. The danger begins only when you change what you do. So before any action — stopping or starting a medication, taking a substance, delaying care you could get — ask one question: what happens if this answer is wrong? If the cost of being wrong is small, you can afford to try. If the cost is your health, the answer is not enough on its own, no matter how confident it sounded. Never stop or change a prescribed medication on an AI's word alone.

This is not a warning against acting. People without access to care must make decisions; that is the whole point. It is a method for making them with your eyes open.

Step 7

Keep what you learn.

Save the conversation. Note what you asked, what you learned, and what changed. Over weeks, those notes become something most patients have never had: a record of your own health, in your own words, organized around your own questions. And if you do eventually reach a clinician, bring it. You will get more from a fifteen-minute visit than most people get from three.

Every step on this page is one piece of a larger discipline we call CLAIM: grounding answers in your context, interrogating instead of accepting, comparing across sources, judging before acting, and carrying each lesson into the next question. The full version is coming as The Practice. For ready-made messages you can copy and send right now, go back to Start here.

You have the right to understand your own health. Start.