PeakRoutine
Personalized health coaching powered by your biomarkers
524 followers
Personalized health coaching powered by your biomarkers
524 followers
PeakRoutine connects your sleep, sunlight, exercise, calories, nutrition, mood, hydration, and more — correlates them against each other — then tells you exactly what it means for your body. No generic plans. Just a proactive AI coach that learns your biology and builds habits around it.
How does PeakRoutine distinguish between a real causal signal and a random correlation? What evidence shows the recommendations actually improve outcomes?
PeakRoutine
Love this question — it's the exact trap we built the whole thing to avoid.
We only surface a pattern when it keeps repeating over time — that's what separates a real signal from a one-off coincidence. And even then it's shown as "likely," never "this caused that." It's always *your* data vs your own baseline, not averages from other people.
Here's the part that matters most though: we don't just tell you, we test it. Suggest one small change, then check whether your numbers actually move. That per-person loop is the strongest kind of proof there is — it's evidence from *your* body, not a "studies say" average that might not apply to you at all.
Do we have formal clinical studies behind it yet? Not yet, and we won't pretend to. But the whole engine is built to earn trust one validated change at a time — which beats confident-sounding guesses every day.
Put it to the test and let it prove itself 🙏
ChatWebby AI
Finally an app that turns my wearable data into something I can actually act on instead of just staring at. Does the habit engine adapt on low-recovery days, or hold me to the same plan?
PeakRoutine
@zain_sheikh This made our day 🙌 — "act on it instead of staring at it" is literally the whole reason we built it.
And yes, it adapts. Holding you to the same plan on a low-recovery day is exactly the streak-guilt trap we wanted to kill. When your recovery, sleep or HRV say you're running low, the engine dials the day back — swaps the hard session for a walk or mobility, protects your streak instead of breaking it, and nudges the harder stuff to a day your body can actually take it.
And it doesn't just hand you tasks — every step tells you what it affects and how it'll help (the "why it matters" + expected impact on your numbers), all laid out as a priority list so you know exactly what to do first instead of guessing.
The goal is consistency with your biology, not in spite of it — push when you've got capacity, recover when you don't, and always know which move matters most.
ChatWebby AI
The correlation engine is the part that stands out to me, surfacing relationships across sleep, HRV and sunlight is genuinely hard to do well at n=1. How many days of synced history do you typically need before the AI starts surfacing confident patterns rather than noise?
PeakRoutine
@zain_sheikh Appreciate that — the n=1 correlation problem is honestly the hardest part to get right
Real answer: it's recurrence-driven, not a fixed day count. We backfill ~90 days the moment you sync, so a lot of the first patterns show up on day one. After that, a high-frequency daily signal (sunlight → that night's sleep) firms up in ~2–3 weeks since you get a data point every day; rarer behaviors take proportionally longer.
So ballpark: confident single-link patterns in a couple of weeks (or instantly if your history already shows them), multi-variable stuff over a month or so. And anything that hasn't earned confidence stays "noticing," not advice — we'd rather under-claim than surface noise.
The recovery-aware habit engine feels like the strongest part. Wearables already show what happened, but turning that into small actions your body can actually handle is much harder. Curious how quickly the routine adapts after a few bad nights.
PeakRoutine
@farrukh_butt1 You get it — any band can tell you that you slept like trash. The annoying part is it still expects you to crush the same workout. That's the bit we built around.
It reacts the next morning, not next week. One rough night, it takes the edge off the day. A few in a row, it properly backs off — hard session → a walk or mobility, gentler targets, and it nudges you to recover instead of guilt-tripping you. It follows the trend, not one reading, so no overreacting to a single late night — and the second you bounce back, it ramps you up again instead of babying you. The plan moves with you instead of making you feel like you failed it.
PeakRoutine
@felix_masera Thanks so much — you've nailed exactly what we set out to do. Most apps personalize on intent ("I want to lose weight"), we personalize on state ("your HRV dropped 18% this week and your sleep efficiency is at 71% — here's what that means for today").
Out of the box we pull from Apple Health, which means anything feeding into it works — Apple Watch, Oura, WHOOP, Garmin, and manual logs for nutrition, mood, and hydration. We're hardware-agnostic by design so you're not locked into one ecosystem.
Direct integrations (WHOOP, Oura Ring, AmazFit etc) are next on the roadmap. Would love to hear what you're tracking — always shapes what we prioritize! 🙏
@siddhant_gupta12 Apple Watch user here — mostly sleep and HRV. The Apple Health-first call is the right one, locking people into one ring/strap from day one is what kills adoption in this category. Going to try the daily summary this week, the "state vs intent" framing alone is worth the install.
PeakRoutine
@felix_masera 100% agree — and honestly, that was one of our earliest and most important product decisions. We kept seeing people drop off health apps not because they lost motivation, but because the app only worked with their specific device. Apple Health as the data layer means you bring your own setup — Watch, Oura, WHOOP, whatever — and PeakRoutine just works.
Sleep + HRV is actually one of the richest signal combinations we work with. HRV tells you how recovered you are, sleep tells you why — together they paint a surprisingly clear picture of what your body is ready for that day.
Would love to hear what you think after your first week of daily summaries, and feedback/suggestion on the app.
PeakRoutine
@sakshi_patil8 Thank you so much ✨ Both great questions:
On your data — it's encrypted in transit and at rest, never sold, and never shared with advertisers. You stay in control: you choose what syncs from Apple Health, and you can export or delete everything whenever you want. It's your health data, full stop.
On personalization — it's the opposite of generic. Every recommendation is built off your baselines and your patterns, not population averages. So instead of "everyone should get 8 hours," it's more like "your deep sleep drops on the nights you train after 7pm — here's what to try." The notifications work the same way: they fire off your actual trends and recovery, so they show up when they're relevant to you, not on a generic schedule. The whole point is advice that sounds like it's about you — because it is.
personalized health coaching from biomarkers is the kind of thing that only works if the recommendations stay coherent over time. how do you handle the case where a biomarker shifts (sleep, HRV, fasting glucose) and your last week's plan suddenly contradicts this week's? does it explain the why behind the change or just hand you the new plan?
PeakRoutine
@thenameisarian This is a sharp question — and one of the core product challenges we've been obsessing over.
The short answer: we don't just hand you a new plan. Every recommendation has two elements baked in — why that step was recommended and its expected impact. Beyond that, when a plan changes, the coaching explains the delta: what shifted in your data, why it matters biologically, and why the recommendation is evolving as a result.
Here's how it actually works in practice:
When a biomarker shifts (say your HRV drops 15% over 5 days while sleep duration looks fine), the system doesn't treat it in isolation. It looks at the cluster — did workout intensity spike? Is sunlight exposure down? The coaching narrative is built around that pattern, not the single number.
The "why behind the change" isn't optional — it's the core UI moment. We found in early beta that users ignored recommendations they didn't understand, even good ones. So the coaching surfaces something like: "Your HRV dropped after you hit 3 high-intensity days in a row — your nervous system is signaling it needs recovery, not another push day." The plan change follows from that explanation, not the other way around.
On coherence over time: this is genuinely hard. We maintain a rolling context of your recent biomarker trajectory so a new recommendation doesn't contradict last week's without acknowledging it. If we told you to prioritize Zone 2 cardio last week and we're now saying dial back intensity, the coaching bridges that — it doesn't just silently flip.
What we're still building: a more robust "plan memory" layer that surfaces when the AI updates its model of you and why — so you can see your health narrative evolve, not just get a new set of instructions every Monday.