Chat · Setup in 10 min· Free taster
Set up ChatGPT for personal health in 10 minutes
A safe, useful baseline — no medical claims, no chatbot diagnosis.
Most people open ChatGPT, paste a symptom, and accept whatever comes back. That's how you get bad advice fast. The setup below makes it useful for reading your own data, not for diagnosing you.
Before you start
- A free ChatGPT account
- One health document you actually want to understand (a lab PDF, a wearable export, a clinician note)
- 10 quiet minutes
The steps
- 01
Turn on Custom Instructions
Settings → Personalization → Custom Instructions. In the first box, paste a one-paragraph self-description: age range, conditions, medications, allergies, the goal of these conversations. In the second box, paste the response rules below.
- 02
Paste these response rules
"Always cite sources when making a claim. Distinguish between observational data, RCT evidence, and expert opinion. Never diagnose. Flag when something belongs in front of a clinician. If you don't know, say so." That's it. No fancier prompt engineering needed for a careful baseline.
- 03
Upload your first document
Drag in the lab PDF or wearable export. Ask: "Summarize what's outside the typical reference range and explain why each one matters in plain language." You're using it as a reading partner, not an oracle.
- 04
Save the thread as your ledger
Rename the conversation "Health Ledger — [your name]". Keep returning to the same thread for follow-ups. The 1M-token context window is the whole point — fragmentation across new chats is how people end up with contradictory advice.
Honest note
ChatGPT will still occasionally hallucinate. Treat anything it says like a research lead, not a verdict. The 10-Day Challenge walks you through the verification habit that makes this safe.
Want the whole stack, not just one tool?
The free 10-Day Challenge wires these together. Or join the free 45-min live workshop and watch me build it end-to-end.