Three free tools.
One stack.
No new app. No subscription you didn’t want. We turn Perplexity, Gemini, and ChatGPT into one specialized health-intelligence stack you operate yourself.
Watch it assemble
Not an app install. An Atlas assembly.
A health operating system you compose from tools you already trust — and stay in charge of.
The shift
From 14 tabs to one biological narrative.
Old standard
Blank-slate patient
Generic content. Fragmented apps. Manual synthesis. Most quit by Week 2.
The reality
AI-informed patient
Patients arrive half-diagnosed, slightly skeptical, highly aware. Helpful tips aren’t enough anymore.
The new mandate
Intelligence-armed patient
One unified thread. 1M+ token memory. Patterns emerge from your real life, not someone’s algorithm.
The architecture
The 3-Layer Stack.
Research mechanisms. Log daily patterns. Synthesize sustainable protocols. Each layer uses a free chat tool with a precise role. Together they form one health operating system — and you stay in charge.
01 · Research
Understand mechanisms.
Tool: Perplexity
The research layer bridges clinical reality with cultural wisdom. Citation-backed synthesis instead of Google's SEO spam. You learn to assess evidence strength — strong RCTs over influencer claims.
Output → A living, personalized reference library.
- Magnesium + Statins: 2hr gap required.
- Ginger + blood thinners: caution / monitor.
- B12 + PPIs: low absorption risk.
- GLP-1 (Mounjaro) + glycinate: take 3hrs apart.
In practice
02 · Ledger
See your patterns.
Tool: Gemini
60-second daily logs in plain language. The 1M+ token memory window sees patterns across weeks, not just days. Continuity over perfection — most quit by Week 2 chasing perfection. The 3x weekly rhythm peaks at Week 3+.
Output → A chronological, synthesized biological narrative.
- What: Had the chicken & rice bowl for lunch.
- How you felt: Energy crashed hard around 2pm.
- Anything different: Slept only 5 hours last night.
In practice
03 · Protocol
Build the plan.
Tool: ChatGPT
The Reality Filter. 90% of generic protocols fail the feasibility test — travel, work, family, GI side effects. The protocol layer binds clinical research to your actual constraints, producing plans you can keep.
Output → A living plan ready for practitioner review.
- Generic: Gym 5x/week, 400mg AM Magnesium → ignores travel.
- Week 2 protocol: 15min walk post-meal, magnesium at 9pm.
- Buffer this sushi craving to keep glucose variability < 12%.
In practice
Run it yourself
The whole method is four moves.
Not a course you watch — a loop you run. Once a week is enough. Each pass takes minutes and the system you build is yours to keep.
- 01
Connect what you already track
Point the stack at the data your phone and wearables already collect. Nothing new to download.
- 02
Run the Ledger prompt
One paste, plain language. The long-context memory starts building your baseline.
- 03
Read your own signals
Patterns get named and mapped — by you, not by an algorithm you can't inspect.
- 04
Decide and adjust
Turn the read into a protocol that fits your real week. Keep what works, retire what doesn't.
That's the loop. The tools change, the AI gets better — the four moves stay the same, and the judgement behind them stays yours.
Where AI fits
The information hierarchy.
Most people jump from Google straight to a practitioner and wonder why they can’t find a good coach. The AI Stack is the missing translation layer.
- 01
Layer 1 · Google
Fast, generic, ad-driven. Returns links, not answers.
- 02
Layer 2 · AI Health Stack You build this
Synthesized, personalized, owned. You ask the right questions and build understanding over time.
- 03
Layer 3 · Practitioner
Human expertise, nuanced refinement, medical care. Now they receive informed context, not vague complaints.
Engineering health prompts
The prompt anatomy.
AI engineered for health requires rigid, multi-variable constraints. We teach a repeatable pattern: [Biological Baseline] + [Real-Time Data] + [Constraints] = [Actionable Output].
Component
[Biological Baseline]
Age, visceral fat, allergies, current meds (e.g., GLP-1).
Component
[Real-Time Data]
HRV / stress meter, sleep score, 24-hour food log.
Component
[Constraints]
Travel schedule, harm-reduction strategy, food sequencing, somatic reset.
Example output card
“Buffer this sushi craving to keep glucose variability below 12% without GI side effects.”
The intelligence sequence
From passive tracking to active pattern recognition in 12 days.
Days 1–3
You log. We decode.
Paste food, sleep, labs. AI reads all.
Days 4–7
Patterns emerge.
Day 6: your biological patterns named & mapped.
Days 8–10
Protocols tested.
Built for your schedule, not a textbook.
Days 11–12
Your own system.
Move into Notion. Yours forever.
See the method in your own life.
The 10-day challenge is the same architecture, condensed. Free, no credit card.