AI reads. Offline data verifies.

Read the prescription.
Capture the consultation.
Never guess.

Two tools for the clinic desk. One reads a handwritten prescription and checks every medicine against a local drug list. The other turns a recorded consultation into a clinical note and a summary the patient can actually read.

Analyse a prescription Record a consultation
  • 254kIndian medicines checked offline
  • 4languages for patient summaries
  • 0recordings kept on the server
Strict, no-guessing OCR
Drug check runs offline
Unclear text is flagged, not filled in
Consent before every recording
Two tools

Built for the two moments that eat up a clinic's time

The prescription is read once and checked against real data. The consultation is captured once and written up for both the doctor and the patient.

Prescription reader

Upload a photo of a handwritten or printed prescription. The text is read exactly as written, then every medicine is matched against a local dataset of 254,000 Indian brands and generics.

  • Raw OCR text is copied exactly. Corrections live in a separate, labelled suggestion field.
  • Brand and generic matching with typo tolerance, entirely offline.
  • Dose, frequency and duration parsed into a plain schedule.
  • Duplicate medicines and low-confidence reads are flagged for a human to check.
  • One-click PDF of the verified result.
Analyse a prescription

Consultation notes

Record the visit in the browser, or upload an existing file. The audio is transcribed with speaker labels, the clinical facts are pulled out, and the patient gets a summary in their own language.

  • Doctor and patient consent is captured before the microphone turns on.
  • Audio is saved on the device every second, so an interrupted visit is never lost.
  • Malayalam, Tamil, Hindi and English, with speaker-separated transcripts.
  • Structured clinical facts plus an after-visit summary for the patient.
  • Nothing is stored on the server once processing finishes.
Record a consultation
How it works

Three steps, and the AI only does the first one

  1. 1

    Read

    Gemini reads the image or listens to the audio and returns the text exactly as it appears. It is not asked to decide what a medicine is.

  2. 2

    Verify

    Every medicine name is matched against the local drug dataset with deterministic, offline code. A match is a fact. A near-miss is a suggestion.

  3. 3

    Review

    Verified items are shown in green. Anything unclear is flagged for you to confirm before it goes anywhere near the patient.

Why it's safe to use

One rule: never present a guess as a fact

Clinical tools earn trust by being honest about what they do not know. Every part of the workspace is built around that.

AI reads, code decides

The model is used only to read text. Whether a drug is real, and which one it is, comes from a dataset on disk. No network call can change that answer.

Raw text is never overwritten

What was written on the page is shown as written. Any AI-corrected name sits beside it, clearly labelled as a suggestion.

Low confidence is visible

Blurry, cropped or ambiguous reads are surfaced for manual review instead of being quietly completed.

Recordings stay yours

Audio lives on the device until you press Process. It is sent once for transcription, then discarded. The server keeps nothing.

Ready when the next patient is

No account, no setup. Open the workspace and drop in a prescription or press record.

Analyse a prescription Record a consultation