DrPenDown, explained.

1. Click New consultation on the dashboard.
2. Type the patient's name. Pick them from the dropdown if they've been here before.
3. Pick a template from the row of pills near the top. Default works fine.
4. Tab through vitals, complaints, exam, diagnosis. Fill what you have, skip the rest.
5. Add one or more medicine rows. Type the brand. DrPenDown suggests from your library first, then a 50k-term drug list. Press Tab to accept.
6. Press CtrlEnter to print. Hand the A4 to your patient.

• Not a medical device. No diagnosis, no clinical decision support, no drug-interaction checks.
• Not an inpatient EMR. No admissions, no bed census, no IPD charting.
• Not an HIS. No lab, pharmacy, billing, or HL7 integration.
• Not a mobile app. Desktop browser only today.

DrPenDown generates a short ID like PID-007 from your initials and the next free number. New patient? Just type the name and tab through. Returning patient? Pick from the dropdown as you type.

DrPenDown saves every 10 seconds while you type. The pill in the top-right shows the live state: Saving… → Saved · 12s ago.

Drafts don't print. They sit in your dashboard until you finalize or press print.

Switch tabs or close the browser mid-consultation? DrPenDown flushes the in-progress note to the server before the tab goes away.

BP (systolic / diastolic), pulse, temperature, weight, SpO₂. Any subset saves cleanly. A patient with only BP and weight is fine. No need to fill all six.

Every clinical text field on the consultation page has a small Mic button at the bottom-right corner. Click it once to start recording. Talk the way you always talk: abbreviations, drug shortcuts, half-sentences, your accent. Click the button again (or wait) to stop.

The mic has four visual states:

• Dictate. Idle. Click to start.
• Stop · 0:0X. Recording (red). Timer ticks every quarter-second.
• Transcribing…. Audio is uploading to Whisper. Usually 1–3 seconds.
• Rewriting…. Spoken commands like "comma" or "new line" are being converted to actual punctuation. Usually under a second.

Hard cap is 60 seconds per recording. The timer turns amber at 50s as a heads-up. If you hit the cap, the recording stops automatically and whatever you've said so far is transcribed.

Only one recording at a time, anywhere on the page. If you click the mic on field B while field A is still recording, B refuses with a toast. Stop A first so you don't lose 30 seconds of dictation.

Every phrase you speak on a field is saved against that field — chief complaints, exam findings, advice, anything you dictate. That's your personal library: a per-field memory of how you write.

The next time you start typing on the same field, autocomplete finishes your sentences from your own history. Dictate "fever, cough, three days" on the chief-complaints field today, and tomorrow the same phrase is at the top of the suggestion list for every new patient.

Your library is also sent to the rewrite model as a glossary. When you say "PCM", the model writes "PCM" — not paracetamol, not acetaminophen, not whatever it would default to. The rewrite is grounded in your words, not in a generic drug dictionary.

DrPenDown's rewrite model only knows two things: what you just said, and the phrases you've already dictated on this field. That's it. There is no general drug knowledge in the loop.

Practical effect: if you've never dictated "linagliptin", the model will not invent "linagliptin" into a transcript that said "linaclotide". It picks up only what your voice produced. Your abbreviations stay yours. Your shorthand stays yours.

As a defense layer, every rewrite passes through a safety check before it lands in the field. If the rewrite drops words from the transcript or invents clinical content, DrPenDown falls back to the raw transcript instead. The cost of a hallucinated clinical note is too high to trust any single model output.

Dictation uses your browser's microphone. On first use, the browser will prompt you for permission. Click Allow. If you clicked Block by mistake, fix it in your browser's site settings.

Mic access requires a secure context. That means HTTPS on the live site, or localhost in development. Plain HTTP deployments won't be able to dictate.

Supported browsers: Chrome, Edge, Brave, Arc, Safari (16.4+). The recorder picks the best audio format the browser supports (Opus in WebM where available, AAC in MP4 on iOS Safari).

If dictation isn't working, the mic button shows an error pill with the reason. Common ones: Microphone access denied, No microphone found, Microphone requires HTTPS or localhost.

Each row has: type (Tablet, Syrup, Capsule, etc.), brand, chemical, dose, frequency, timing, duration, instructions. The order matters less than the order you type them in.

Start typing the brand. DrPenDown ranks suggestions in this order: your own library (by use-count) → 50k-term drug dictionary.Tab accepts. ↓ / ↑ walks the list.

Every prescription you save is added to your My Rx library. Grouped by brand. Search across brand + chemical. Sort by "most prescribed" or "most recent". Edit any format inline.

Next consultation: the same brand + dose + frequency autocomplete from your own history. The system gets faster the more you use it.

CtrlEnter prints the current consultation. DrPenDown renders the A4 PDF and opens your printer dialog. Your clinic letterhead is at the top. Logo, name, qualification, registration number.

Settings → Branding. Fill in: clinic name, address, phone, email, website, qualification, registration number. Upload a logo (≤ 1 MB, PNG / JPG / SVG).

Drag the on-screen positioner to move the logo. Every PDF picks up these details automatically.

A template is the list of sections DrPenDown renders on every consultation: Chief Complaints, HPI, Past History, Exam, Diagnosis, Plan, and so on. DrPenDown ships with sensible defaults. Clone one, reorder sections, rename, and you have your own.

Open Templates. Create consultation templates or discharge templates. Clone an existing one to start. Up / down arrows reorder sections. Delete the ones you don't use.

There isn't an in-app upload button yet. Import runs through a small CLI that talks straight to your Convex deployment. It accepts exports from Practo, Lybrate, DocOn, Bahmni, and any generic EMR.

From a checkout of the codebase:

node scripts/import-csv.mjs patients ./data/patients.csv
node scripts/import-csv.mjs visits   ./data/visits.csv
node scripts/import-csv.mjs media    ./data/media.csv

Add --prod to push to the live deployment instead of local. Schema: see convex/import.ts.

Send us a sample row at support@drpendown.com. We'll write a one-off importer. Usually a same-day turnaround.

An in-app export button for consultations, patients, and templates isn't available yet. Today, the same data is reachable from the Convex dashboard (per-table JSON view) or on request. Email support@drpendown.com and we'll send a JSON or CSV dump of every table you own.

• Command palette / searchCtrlK
• Save current consultationCtrlS
• Print current consultationCtrlEnter
• Toggle sidebarCtrlB
• Show this panelCtrl/

• DashboardAltD
• New consultationAltN
• DictionaryAltB
• The other nav items: Consultations, My Rx, Templates, Patients. None of them have shortcuts yet. Use CtrlK to jump anywhere via the command palette.