Research
How the ATLAS composition came about
This article does not name a single ingredient in ATLAS. It is about the question before that: how the selection came about - and how a selection became a dose.
What follows is the procedure: which questions are put to every amount, in which order - and what makes a proposal fail, including one of our own.
Behind a dose lie many decisions the number never shows.
The first question is not whether an amount is safe
That an amount is safe does not yet mean it makes sense. So for every substance we first looked at how the effect changes in studies as the amount goes up. There are 4 cases.
4 shapes a dose-response relationship can take. Horizontally the amount, vertically the benefit.
- linear - more is more, no peak has been found.
- plateau - above a certain amount, more adds nothing. The most common case.
- optimum - above a certain amount, more gives less again.
- unknown - there is no dose-response analysis.
The technical term for the third case is biphasic: a response that goes one way at a low amount and the other way at a high one.1,2 In an analysis of 26 randomised trials (comparison studies in which chance assigns the groups) with 1,831 participants, the largest effect on systolic blood pressure lay between 100 and 200 mg per day; above that it decreased again.3
The questions put to every amount
Every ingredient in the formula had to answer these questions, one by one.

First we look at what happens in studies when the amount goes up. Does the effect keep growing? Does it stop at some point? Or does it shrink again? Only that tells you which range makes sense.
It depends on what the study measured. One and the same substance was studied twice: once for blood pressure, once for markers of oxidative stress (strain on cells from reactive oxygen compounds). For blood pressure the best range was 100 to 200 mg a day,3 for oxidative stress 100 to 150 mg.4 So there is no single right amount.
Some limits are law, others only a recommendation. For one substance an EU regulation sets 150 mg per daily portion - more may not be sold.5 For other substances an authority only names a recommended maximum. We kept the 2 apart.
A supplement does not have to deliver 100 % of the recommended daily amount. Most of it comes from food. So for every substance we also counted what the diet adds. For one trace element a proposal set 15 mg. Men in Germany take in 11.6 mg from food at the median.6 Together that would have been 26.6 mg - more than the European upper limit of 25 mg.7 The proposal was dropped.
An upper limit is meant for a whole day, not for a single moment. The European value for one mineral comes from studies in which the amount was spread across the day.7 4 capsules once a day are a single intake. We took that into account and stayed below.
For an ingredient it matters which nutrient compound it comes in and what it is bound to. That affects, among other things, how well it is tolerated and how much of it arrives in the body. Between 2 preparations of the same substance this can differ manyfold.8
A limit does not apply to everyone alike. The European upper limit for one nutrient is 1,000 µg a day for adults and 800 µg for 15- to 17-year-olds.9 A national proposal works with the lower figure.10 We defined who the product is for and calculated accordingly.
4 capsules are a hard limit. A capsule holds about 700 mg, which makes about 2.8 g a day - for everything together. Substances given in grams do not fit: for one mushroom extract the daily amounts studied ranged from 1 to 3 grams.11,12 Collagen and creatine therefore could not be included.
What counts is not only whether an amount is permitted, but whether there are warning signals. For one plant preparation no safe intake level could be derived in 2018: from 800 mg a day liver values rose in studies, and harm could not be ruled out below that either.13,14 Where no safe amount can be named, what decides is how far below it you stay.
2 ways to an amount
For every amount there are 2 defensible ways. Either you stay inside the range in which the substance was studied - or you deliberately give little, but daily over years. For the second way most of these substances have not been studied at all. So we claim no effect, we only describe how the decision was made.
Across more than 30 amounts examined, a substance dropped out entirely for 4 reasons: the amount is many times below anything ever studied. The raw material does not state how much of the active constituent it contains - then no study fits it. Studies exist and they argue against it.11 Or the substance carries conditions that do not fit daily intake.
None of these 4 reasons is a verdict on the substance. It is a verdict on its place in this particular product.
We turned the tables
For a substance to be included, it had to withstand the best arguments against it - and those were written down first, not afterwards.
That is the devil’s advocate’s job: to argue the other side as seriously as you would want to be convinced yourself. Only what survives that stays in.
Looking at other products belongs to this too, but never as proof: the question is not what others have, but whether we would propose the same if nobody had it. That comparison produced 30 suggestions for changing our own composition. 25 did not survive the counter-check. 5 remained.
The rules the decisions followed
How much work went into a composition says nothing about how good it is. What counts is how the work was done. These rules applied to every single substance, without exception.
We asked the same questions about every substance, in the same order. The first one is not “is this safe”, it is “how does the effect change as the amount goes up”.
Every figure comes from the authority’s original opinion. Not from a table summarising it - these values change. One European upper limit was lowered in 2023 from 25 to 12 milligrams a day.15
We kept law and recommendation apart. A regulation is complied with. With an authority’s proposal, going above or below it has to be explainable.
Every result was checked a second time. That second check looks for the mistake in our own result, not for confirmation.
Every substance comes with the counter-question. What argues against our own recommendation - and what exactly does the work that argues against it say?
We compared the ingredients and doses of a great many competitors. Not to copy, but to see what other teams had worked on, to have no blind spots, and to be able to place our own product concretely.
Our own knowledge section is a reference work. Every article there is researched and names its sources, mostly primary papers. So when a question comes up, that is where we look first.
When a figure changes, it changes everywhere. Label, website and data sheet are checked against each other. Whoever looks something up should find the same figure - no matter where they look.
At the end we had everything counter-checked by artificial intelligence. Not to hand over decisions, but to find blind spots: questions we would not have thought of ourselves. A human made the final call.
What decides in the end - and who
I am a specialist in surgery and first put this composition together for myself. I am, so to speak, the first customer. That explains why single milligrams are argued over here.
A procedure does not take the decision away from anyone. It only makes the decision traceable. Where the evidence is not enough, nothing goes in. Where it speaks against an ingredient, that ingredient goes - even if it would have sold well. I made these decisions.
What this article does not claim
This article says nothing about what a single substance does in the body. For most of the substances examined there is no health claim authorised in the EU. Where there is none, we may not claim anything - not even with a source.
It says nothing about other companies’ formulations either. And every figure in it has a source you can look up.
A current state, not an end point
What comes out of this procedure is a current state. Upper levels are re-derived, and a decision that is well argued today can be out of date in 2 years. Then we change it - the same way. How we check is set out under Science.
Key takeaways
- What counts first is how the effect changes with the amount, not the size of the number.
- Every amount is checked from several directions: curve, measure, legal status, sum with the diet, serving, formulation, target group, mass and warning signals.
- A legal maximum and an authority’s proposal are 2 different things.
- 2 defensible ways lead to an amount: stay inside the range studied - or deliberately give little, but every day. A substance drops out entirely when the amount is far below any study, when the raw material does not state its content, when studies argue against it, or when it carries conditions that do not fit daily intake.
- For every proposal the counter-thesis was written down and tested. Only what survived it remained.
- At the end artificial intelligence counter-checked to find blind spots - a human made the decision.
- The result is a current state, not an end point.
Sources
- Calabrese EJ, Baldwin LA. Defining hormesis. Hum Exp Toxicol. 2002;21(2):91-97. PMID 12102503.
- Mattson MP. Hormesis defined. Ageing Res Rev. 2008;7(1):1-7. PMID 18162444.
- Zhao D, Liang Y, Dai S, et al. Dose-Response Effect of Coenzyme Q10 Supplementation on Blood Pressure among Patients with Cardiometabolic Disorders: A GRADE-Assessed Systematic Review and Meta-Analysis of Randomized Controlled Trials. Adv Nutr. 2022;13(6):2180-2194. PMID 36130103.
- Dai S, Tian Z, Zhao D, et al. Effects of Coenzyme Q10 Supplementation on Biomarkers of Oxidative Stress in Adults: A GRADE-Assessed Systematic Review and Updated Meta-Analysis of Randomized Controlled Trials. Antioxidants (Basel). 2022;11(7):1360. PMID 35883851.
- European Commission. Commission Implementing Regulation (EU) 2021/51 of 22 January 2021 authorising a change of the conditions of use of the novel food ‘trans-resveratrol’. OJ L 23, 25.1.2021, p. 10.
- Max Rubner-Institut. Nationale Verzehrsstudie II, Ergebnisbericht Teil 2. Karlsruhe, 2008, p. 141 (median zinc intake).
- Scientific Committee on Food / EFSA NDA Panel. Tolerable Upper Intake Levels for Vitamins and Minerals. European Food Safety Authority, February 2006 (zinc p. 197; magnesium pp. 112-113).
- Riva A, Ronchi M, Petrangolini G, et al. Improved Oral Absorption of Quercetin from Quercetin Phytosome®, a New Delivery System Based on Food Grade Lecithin. Eur J Drug Metab Pharmacokinet. 2019;44(2):169-177. PMID 30328058. Manufacturer study, 12 subjects.
- EFSA NDA Panel (Turck D, Bohn T, Castenmiller J, et al.). Scientific opinion on the tolerable upper intake level for folate. EFSA J. 2023;21(11):e08353. PMID 37965303.
- German Federal Institute for Risk Assessment (BfR). Höchstmengenvorschläge für Folsäure in Lebensmitteln inklusive Nahrungsergänzungsmitteln, Aktualisierung 2024 (in German). Opinion No. 009/2024 of 22 February 2024.
- Parcell AC, Smith JM, Schulthies SS, et al. Cordyceps Sinensis (CordyMax Cs-4) supplementation does not improve endurance exercise performance. Int J Sport Nutr Exerc Metab. 2004;14(2):236-242. PMID 15118196.
- Chen S, Li Z, Krochmal R, et al. Effect of Cs-4 (Cordyceps sinensis) on exercise performance in healthy older subjects: a double-blind, placebo-controlled trial. J Altern Complement Med. 2010;16(5):585-590. PMID 20804368.
- EFSA ANS Panel (Younes M, Aggett P, Aguilar F, et al.). Scientific opinion on the safety of green tea catechins. EFSA J. 2018;16(4):e05239. PMID 32625874.
- European Commission. Commission Regulation (EU) 2022/2340 of 30 November 2022 amending Annex III to Regulation (EC) No 1925/2006 as regards green tea extracts containing (-)-epigallocatechin-3-gallate. OJ L 310, 1.12.2022, p. 1.
- EFSA NDA Panel (Turck D, Bohn T, Castenmiller J, et al.). Scientific opinion on the tolerable upper intake level for vitamin B6. EFSA J. 2023;21(5):e08006. PMID 37207271.
ATLAS All-in-One - 35 ingredients, 4 capsules, once a day.
