How RXID.ai Identifies Medications

Five identification methods powered by AI and federal drug databases. Here's how each one works.

Text Search

Type a drug name, generic name, or the imprint code printed on the pill. RXID queries the RxNorm API from the National Library of Medicine to match your search against standardized drug names. Results include brand names, generics, NDC codes, dosage forms, and manufacturer details.

RxNorm normalizes different representations of the same drug. So whether you search "Lipitor," "atorvastatin," or "atorvastatin calcium," you'll find the right result.

Photo Upload with AI Vision

Upload a photo of any pill, tablet, or capsule. RXID sends your image to GPT-4o, OpenAI's multimodal vision model, which analyzes visible characteristics: color, shape, size, markings, scoring lines, and coating type.

The AI generates a structured identification including probable drug name, dosage, manufacturer, and confidence level. Results are cross-referenced against FDA's NDC directory for verification.

Live Camera Capture

On supported devices (most modern smartphones and tablets), RXID accesses your rear camera for real-time pill capture. Position the pill inside the viewfinder frame and tap the capture button. The captured frame is processed identically to an uploaded photo.

On devices without camera API support, RXID automatically falls back to your device's native camera app with a "Take Photo" button.

Barcode and NDC Scanning

Pharmacy bottles and medication packaging include barcodes that encode National Drug Code (NDC) numbers. RXID's barcode scanner reads Code128, UPC-A, and EAN-13 formats, extracts the embedded NDC, and looks it up against FDA's NDC directory.

You can also type an NDC manually if the barcode is damaged or unreadable. NDC formats (4-4-2, 5-3-2, 5-4-1, 5-4-2) are all normalized automatically.

Attribute Search (Color, Shape, Imprint)

When you can read text on a pill but don't have a clear photo, use attribute search. Select the pill's color from 12 FDA standard colors, choose its shape from 10 FDA standard shapes, and enter any visible imprint text. RXID queries FDA's openFDA database and returns matching medications with C3PI pill images from NIH.

Data Pipeline

Every identification method connects to the same trusted backend:

  • NIH/NLM: RxNorm for drug name normalization, C3PI for pill reference images, DailyMed for drug labeling
  • FDA: NDC directory for product identification, openFDA for drug labels and safety data
  • GPT-4o: Visual analysis for image-based identification (upload and camera methods only)

Results are cached in Cloudflare's D1 database for fast repeat lookups. All data originates from federal sources and is not user-generated.

Frequently Asked Questions

Use the Attributes tab in the search tool. Enter the exact characters you see on the pill in the imprint field. Pay close attention to similar-looking characters: the letter O versus zero, the letter I versus the number 1, and uppercase versus lowercase. If the pill has markings on both sides separated by a score line, enter both sides. You can also add color and shape to narrow results further. The search checks against the full FDA imprint database.
Most pharmacy pill bottles have a barcode (usually UPC or Code128 format) that encodes the National Drug Code (NDC). The barcode scanner uses your phone camera to read this code, extracts the NDC, and looks it up in the FDA NDC Directory. This gives you the exact drug name, manufacturer, strength, and packaging information. It works best when you still have the original pharmacy bottle. If the barcode is damaged or faded, you can also type the NDC number manually.
Four primary sources. First, the NIH C3PI (Computational Pharmaceutics and Pill Identification) database, which contains reference pill images with verified identifications. Second, DailyMed from the National Library of Medicine, which has FDA-approved labeling and pill descriptions. Third, the FDA NDC Directory, which maps every National Drug Code to its drug product. Fourth, openFDA, which provides drug recall, adverse event, and label data. We also use RxNorm for standardized drug naming across sources.
Yes. The camera feature uses your rear-facing camera to capture a photo directly in the browser. Position the pill on a plain, well-lit surface and tap the capture button. The image is sent to GPT-4o Vision for analysis. On iPhones running iOS 14.5 or later, this works in Safari over HTTPS. On Android, Chrome and Firefox both support it. If your browser does not support camera access, the app automatically falls back to a file picker that opens your native camera app.
Pill Expert AI
Drug Information Specialist
Hello! I can help with medication questions. Ask me about drug identification, side effects, interactions, or dosage information.