AI Pharmacist
Prescription Review on FDA Labeling
Enter an order and the patient facts you have. It resolves each medicine, runs versioned clinical rules against it, and answers drug questions in the label's own words — with the passage, the release and the rule version attached to every finding.
FDA labels ingested
the full DailyMed release
Active ingredients
prescription and OTC
Label sections indexed
individually citable
What it does
Every finding cites a label passage
A finding names the section, the release it came from and a hash of the exact passage. Nothing is asserted that the labeling does not say, and the citation is reconstructible after the fact.
Deterministic rules, not a chatbot verdict
Dose ceilings, renal thresholds, duplicate therapy, allergy cross-reactivity and high-alert combinations are versioned rule artifacts evaluated in code. The model reads and explains; it never decides.
It tells you what it did not check
The corpus covers thousands of ingredients; the clinical rule set covers a reviewed subset. A medicine outside it is reported as unchecked — a review that found nothing is never presented as a clean one.
Missing data changes the answer
Weight, renal function and pregnancy status are tracked as known, unknown, stale or conflicting. A check that needs a fact it does not have reports insufficient data rather than assuming a value.
Advisory, and supervised by a pharmacist
This is decision support, not a decision. Every output requires review by a qualified pharmacist before it informs care. It serves public-domain FDA labeling only — no licensed interaction compendium and no therapeutic guidelines — and questions those would answer are reported as unsupported rather than inferred. The clinical rule set is under review and is not yet approved for clinical use.
Review an order
Open the console — nothing to install, and no patient identifiers are required.
Open the console