Editorial policy.

Sourcing.

Emersus AI's responses are grounded in a continuously updated corpus of peer-reviewed biomedical and exercise-science research. Our primary sources are PubMed, Europe PMC, and open-access journal archives, supplemented with preprint servers (bioRxiv, medRxiv). Every claim in a generated response carries a citation to its underlying source.

Evidence grading.

Claims are tagged by strength of evidence:

• High — supported by multiple peer-reviewed RCTs or a recent high-quality meta-analysis.
• Moderate — supported by a single peer-reviewed RCT or consistent observational evidence.
• Limited — supported by pilot studies, small trials, or mixed observational evidence.
• Insufficient — not enough quality evidence exists to make a confident claim.

Preprint vs. peer-reviewed.

Preprints — research that has not yet passed peer review — are always labeled distinctly. We surface them because they're often the first evidence on a topic, but we flag them so you can weight them appropriately.

AI generation vs. human editing.

Chat responses are AI-generated from the retrieved source corpus. Static site pages (this one, /about/, /privacy/, /terms/, /consumer-health-data/, marketing copy on the landing) are written and edited by humans on the Emersus team.

Updates and corrections.

Static pages show a "Last updated" date. When research evolves or we update the pipeline, we refresh affected pages. If you believe content on this site is inaccurate or outdated, email info@emersus.ai and we will investigate.

Conflicts of interest.

Emersus AI has no sponsors and does not accept payment from supplement companies, pharmaceutical companies, gym chains, or any third party whose products or services could influence what the system surfaces. Recommendations are driven by the evidence corpus alone.

Not medical advice.

See our About page and the health disclaimer in the footer of every page.