The noise and the signal — how AI literacy turns longevity guesswork into a quantified n=1.

You've tried most of it. The cold plunge. The mouth tape. The creatine. The magnesium. The fasting window. The new supplement someone in your feed swears is the one — usually between an unboxing video and a photo of their abs. You feel something. You see something. Sleep is maybe better. Mood is maybe steadier. The mirror is maybe kinder on Thursdays.

But nothing — not Apple Health, not Oura, not your group chat, not the influencer who sold you the protocol — can answer the only question that matters: which of these is actually doing the work?

why the feed will never tell you

Instagram and TikTok university optimise for the next intervention, not the last one. The business model is volume of recommendations, not honest attribution. Every creator has a stack; nobody publishes the counterfactual.

Wearables don't help either. Oura, Whoop, and Garmin can each measure their own slice well, but none of them know that you started creatine on the 11th, switched to mouth tape on the 14th, drank with friends on the 18th, and travelled on the 22nd. The signal is in the joins, and the joins live in your head.

the part you've been missing isn't another app

It's a synthesising layer. Something that holds the qualitative — "slept badly, anxious dream, woke at 4" — beside the quantitative — HRV 38, REM 62 min, step count 4,200 — and can read both at once, across weeks, and tell you what changed.

Until 2023 that layer was a coach, a clinician, or a very patient spouse. From 2024 onward a general-purpose model with a long enough context window can hold it for you. That is the entire premise of AI literacy applied to your own body.

what a quantified n=1 actually looks like

1. Baseline 2–4 weeks of the outcome you actually care about — sleep quality, energy by 3pm, training capacity, mood, recovery. One sentence a day is enough. Wearable summary is a bonus, not the foundation.
2. Pick ONE intervention. Not your whole protocol. One. Creatine. Or magnesium. Or the 9pm cut-off. Or zone 2 four times a week. Run it as a single variable for 4–12 weeks depending on how slow the outcome is.
3. Keep the ledger boring. Date, intervention status, the outcome score, one line of context (travel, illness, hard week). Anything else is noise.
4. At the end of the window, paste the whole ledger into a sourced AI and ask: 'What changed, what stayed the same, where am I fooling myself, and what is the most likely confounder?' Read the answer slowly.
5. Decide: keep, drop, or extend. Then run the next single variable.

this is what AI literacy actually means

It is not prompt-collecting. It is not knowing which model is fastest this month. It is the small, learnable habit of writing your own life down in a format a model can read, asking the model honest questions about it, and judging the answers with grade-aware evidence (Hashimoto et al., 2025).

Done this way, the model is not your doctor and not your coach. It is the synthesising layer that used to be missing — the thing that finally tells you whether the last six weeks of effort moved a number, or whether the algorithm just moved your attention.

where the line is

Diagnosis, prescription, and dose decisions belong with a clinician. Lab interpretation belongs with someone trained to read labs. AI is excellent at synthesising what you already collect into one honest signal, and at preparing you for the conversation with the human who is allowed to make the call.

You do not need another app. You need to escape the algorithm long enough to read your own data. That is the whole job — and it is exactly what the course at /courses, the membership at /membership, and the free 10-day challenge at /start are built to teach.