Ones
One supplement, designed by AI from your blood & wearables
52 followers
One supplement, designed by AI from your blood & wearables
52 followers
Stop guessing with your health. Most "personalized" supplements are just a quiz with a markup. Ones is different: an AI practitioner reads your real blood work, wearable data, goals, medications, and history then designs one bespoke formula, freshly compounded into a single daily supplement made for your exact body. No cabinet of 8–10 generic bottles. No one-size-fits-all green powders. And it adapts as your data changes. One supplement. Actually yours.
Hey Product Hunt 👋 I'm Pete, founder of Ones.
This started with a problem I couldn't let go of: I was spending a fortune on supplements and had no idea if any of it was doing anything. My counter looked like a pharmacy — 8, 10 bottles — all built for some "average" person who doesn't exist. None of it was made for my body.
The deeper I looked, the more absurd it got. We have blood work, wearables, and decades of ingredient research — yet the industry's idea of "personalized" is a 5-question quiz that sells you a slightly different bundle.
So we built the opposite. With Ones, you connect your blood work and wearable data and talk to an AI practitioner about your goals, the medications you take, and your history. It designs a single, fully personalized formula and freshly compounds it into one daily pack — made for your exact body, and it adapts as your data changes.
Honestly, the hard part wasn't the AI — it was the safety and the science. Making sure "personalized" never means "random": real ingredients, safe dosing ranges, and awareness of interactions with what you already take. That's where most of the last two years went.
You can build your own in ~2 minutes → www.ones.health
I'll be here all day. Tear it apart, ask me anything, tell me what's missing — that's exactly what I'm hoping for. 🙏
@peter_stadniuk1 The personalized-supplement angle is interesting, especially if it’s actually driven by lab data rather than a questionnaire. How are recommendations validated and reviewed, does a licensed healthcare professional oversee the AI-generated formulas, particularly when medications, deficiencies, or potential supplement interactions are involved?
@nicole_hynek
Thank you and great questions! Going to give you a super detailed answer on this, hope you don't mind!
Lab-driven, not a quiz. Upload bloodwork PDFs from Quest, LabCorp, Function Health, Inside Tracker (any major lab) and the system extracts the actual markers such as ApoB, Vitamin D 25-OH, Magnesium RBC, Homocysteine, HbA1c, Omega-3 Index, full lipid panel, fasting insulin, etc. Those drive the formula. Same marker in different bodies produces different formulas so ApoB 145 in a healthy 35-year-old leans toward EPA + berberine; the same number in a 65-year-old T2 diabetic leans toward magnesium and glycemic context first.
Deficiencies specifically drive dose ranges, not just ingredient selection. Vitamin D at 22 ng/mL pushes D3 to a specific high-end range; 38 ng/mL pushes it down to maintenance. Every dose decision is calibrated to the severity of the input data, with a built-in 90-day retest expectation formulas update as new data lands.
Validation — 5 layers run server-side before a formula ships:
1. Safety validator hard-blocks ingredients based on your medications list (statins, blood thinners, SSRIs, MAOIs, thyroid meds, etc.), conditions (pregnancy / nursing / specific disease flags), and allergies. Can't be bypassed. For medications the system can't confidently match to its known interaction database, the formula surfaces an explicit caution and recommends physician review before you rely on it.
2. Cross-family AI critic: if Claude generated the formula, GPT critiques it (or vice versa). Different model families don't share blind spots, so the critic actually catches things the generator missed (typically dose-floor violations on the most clinically severe finding).
3. Deterministic dose-band critic against an evidence registry actually flags doses below the primary research band or above the supportive maximum.
4. Selection-gap critic catches "you flagged high ApoB but nothing in this formula actually targets it."
5. Server-enforced ingredient catalog + hard capsule budget. The AI literally cannot pick an ingredient outside the validated catalog or exceed safe dose ranges. Every block + override is logged to an audit trail.
So, as you can see we have a very comprehensive system in place to take into account all of the things you listed out such as deficiencies, medications, etc.
On licensed MD oversight — yes. At this stage of the launch, every formula passes through human clinical review before it ships, on top of the AI safety stack above.