About & Methodology

How we generate recommendations, what the evidence means, and why you can trust us.

Who Built This

Evidence Stack was built by Adam Hultman, a software developer based in Vancouver, BC. After spending too many hours on Reddit threads and YouTube videos trying to figure out which supplements actually had science behind them, he built the tool he wished existed: something that would give straight answers grounded in real research, without trying to sell him anything.

This is a personal project born out of genuine frustration with the supplement industry's opacity. It's not backed by a supplement company. It's not funded by affiliate revenue. It's a tool that tries to give you honest, evidence-based guidance.

How Recommendations Are Generated

  1. 1
    You tell us your goals and context. The wizard collects your health goals, current supplements, age range, sex, medications, dietary restrictions, and budget preference.
  2. 2
    AI analyzes against a curated evidence database. We maintain a hand-curated database of supplements, each with pre-assigned evidence grades, key studies, known interactions, and dosage guidelines. The AI (GPT-4o) selects from this database — it cannot invent supplements or fabricate studies.
  3. 3
    Post-processing for safety and quality. After the AI responds, we run automated checks: deduplication (so you don't get two forms of the same compound), interaction warnings from our safety database, and cost estimates based on typical retail pricing.
  4. 4
    You get transparent results. Every recommendation shows its evidence grade, the specific studies it's based on, known interactions, warnings, and estimated cost. Nothing is hidden.

What Evidence Grades Mean

A

Strong Evidence

Multiple well-designed randomized controlled trials (RCTs) with consistent results across diverse populations. Meta-analyses or systematic reviews support the finding. This is the highest standard of supplement evidence.

B

Moderate Evidence

At least one well-designed RCT, or multiple observational studies with consistent findings. Evidence is promising and generally supports the recommendation, but may have limitations in study design or population diversity.

C

Preliminary Evidence

Limited human studies, often small sample sizes or pilot studies. May include strong mechanistic or animal data that hasn't been fully validated in humans. Promising but not yet robust enough for confident recommendation.

D

Traditional Use Only

Based primarily on traditional or historical use with minimal scientific evidence. May have anecdotal support but lacks rigorous clinical trials. Included for completeness but recommended with significant caveats.

Where Citations Come From

Every study cited in Evidence Stack comes from our curated database of peer-reviewed research. Sources include:

  • PubMed-indexed journals — peer-reviewed medical and nutrition journals (BMJ, JAMA, Nutrients, J Clin Psychiatry, etc.)
  • Cochrane Reviews — systematic reviews and meta-analyses considered the gold standard of evidence synthesis
  • Professional position statements — from organizations like the International Society of Sports Nutrition (ISSN)

We do not cite blog posts, influencer claims, or manufacturer-funded studies without independent replication. The database is reviewed and updated periodically.

No Affiliate Links. No Product Sales. Ever.

Evidence Stack does not sell supplements. We do not have affiliate relationships with any supplement manufacturer, retailer, or marketplace. We never have, and we never will.

This is a deliberate choice. The moment a recommendation engine profits from recommending specific products, its incentives are corrupted. You can't trust advice from someone who gets paid when you buy what they suggest.

Our business model is simple: the free tier gives you solid recommendations. The Pro tier ($15/month) adds advanced features like stack history, full citation links, PDF exports, and priority AI analysis. We make money from subscriptions, not from selling you products.

⚠️ Limitations & Honest Caveats

  • This is not medical advice. Evidence Stack is an educational tool. It does not replace consultation with a qualified healthcare provider.
  • AI can make mistakes. While the AI is constrained to our curated database, it may occasionally misjudge which supplements are most relevant to your specific situation.
  • Evidence evolves. Supplement research is ongoing. A Grade B today may become Grade A (or be debunked) as new studies are published.
  • Interaction data is not exhaustive. Our interaction checker covers common interactions but cannot replace a pharmacist's review of your full medication list.
  • Cost estimates are approximate. Prices vary by brand, retailer, form, and region. Estimates are based on typical US retail pricing and should be used as rough guidance only.
Build Your Stack

Free. No account required. See the evidence for yourself.