About Priolix
Independent, evidence-based analysis of dietary supplements — powered by real research, not marketing.
What We Do
Priolix analyzes dietary supplements sold in the United States by cross-referencing product labels with published scientific research. We evaluate what the evidence actually says about each ingredient, at the doses actually sold — not just what brands claim on the front of the bottle.
Our database covers 170,000+ products from 4,500+ brands, linked to 14,000+ research sources including meta-analyses, systematic reviews, randomized controlled trials, and clinical practice guidelines.
Where Our Data Comes From
DSLD
U.S. NIH Dietary Supplement Label Database — the authoritative source for product labels, ingredient lists, and serving sizes for all supplements sold in the U.S.
PubMed
NLM's bibliographic database of 35M+ biomedical citations. We index meta-analyses, systematic reviews, RCTs, and clinical trials relevant to dietary supplements.
Cochrane Library
Gold-standard systematic reviews and meta-analyses. 725 Cochrane reviews indexed with quality scoring and evidence-level classification.
ClinicalTrials.gov
3,400+ registered clinical trials for supplement interventions, with status, phase, and outcome data from the U.S. national registry.
FDA CAERS
FDA Center for Food Safety and Applied Nutrition Adverse Event Reporting System — real-world safety signals from consumers and manufacturers.
EFSA & ODS
European Food Safety Authority health claim opinions and NIH Office of Dietary Supplements fact sheets — regulatory and educational source material.
What Makes Us Different
- Not an affiliate site. We don't sell supplements or earn commissions. Our analysis is independent of product manufacturers.
- Real evidence, not cherry-picked. We index the full body of research for each supplement — including studies that show no effect or harm.
- Product-level analysis. We don't just review supplements in the abstract. We analyze specific products, at specific doses, against specific evidence.
- Safety scoring. Every product gets a safety score (0–100) based on UL exceedance, unsupported health claims, and ingredient risk factors.
- Open methodology. Our scoring, tiering, and evidence-mapping methods are fully documented on our Editorial Policy page.
By the Numbers
Technology
Priolix is built with Python, FastAPI, and SQLite. Evidence retrieval uses BM25 + FAISS semantic search. AI summaries are generated locally using open-weight language models — no data is sent to third-party APIs. The entire pipeline runs on a single server with GPU acceleration for embeddings and inference.
Important: Priolix does not provide medical advice. See our Medical Disclaimer and Editorial Policy for details.