An eye OPD is one of the fastest-moving rooms in Indian healthcare. Patients move through vision, refraction, and workup counters, then reach the ophthalmologist for a consult that can be over in minutes. AI Medical Scribe is one module inside Practice Copilot, the platform Patient Square builds. It listens during the visit and hands back a structured SOAP note, ICD-10 suggestions, and a prescription draft, ready to review and sign about two minutes after the visit. It hears Hindi, Hinglish, and 20-plus Indian languages; the note always comes out in clean clinical English. Pricing for a practice starts from ₹1,599 per doctor per month on the annual Assist plan, plus 18% GST (about ₹1,887 all-in), with unlimited notes and no per-note metering. A larger eye hospital runs on Hospital Copilot, which can use the complete Patient Square HIS/EHR or work alongside an existing HIS/EHR and is demo-priced.
The bottleneck in a high-volume eye hospital isn’t the exam. It’s what the exam leaves behind.
Key takeaways
- An eye OPD is a flow of counters; the scribe sits at the one that produces the note, the doctor’s chair.
- A structured English note lands about two minutes after each consult, drafted from what the doctor says.
- From ₹1,599/mo per doctor (annual Assist, ex-GST) with unlimited notes and no per-note metering: a 120-patient day costs the same as a 30-patient one.
- It captures spoken findings, not slit-lamp, OCT, or autorefractor output. Those stay on your machines.
- ABDM integration is roadmap, not live, and we say so before you ask.
An eye hospital is a flow of counters, and the note is the last one
Walk a busy eye hospital in Hyderabad or Coimbatore at 11am and you see an assembly line, in the best sense. Registration. Vision and refraction. Dilation and wait. Basic workup, sometimes imaging. Then the ophthalmologist. Each counter has its own queue and its own system, and the whole design exists to keep the doctor’s chair moving.
That chair is where documentation is born and where it backs up. The consult itself can run three to five minutes once the workup is in hand. The note doesn’t move that fast. So at the end of a 90-patient session, the doctor is holding a stack of half-remembered encounters, and the record either gets thin or gets written from memory at 8pm. Neither is good for a chart you may need to defend later.
Picture a cataract camp Tuesday. Two hundred patients screened by lunch, a senior consultant seeing the surgical candidates one after another, three minutes each, calling out lens grades and pressures while the next patient is already being dilated. Nobody is typing during that. The notes get reconstructed after the last patient, or they don’t get written properly at all. That’s the exact gap an ambient scribe fills: it was listening while the doctor talked, so the draft is already sitting there.
The scribe slots into exactly one point in that flow, the doctor’s chair, and drafts the note from the conversation there. The counters upstream don’t change. If your OPD runs 80-plus patients a doctor a day,
A patient-flow documentation map for a high-volume eye OPD
Here’s the artifact: a map of where the patient goes, what each stop produces, and where a written note is actually owed.
| Flow stop | What happens | What it produces | Where the note is owed |
|---|---|---|---|
| Registration | Demographics, complaint intake | Queue token, basic details | Your front-desk system |
| Vision + refraction | Visual acuity, refraction | Numbers on your existing record | Not the doctor’s note |
| Workup / imaging | IOP, OCT, biometry as needed | Device output on your machines | Not the doctor’s note |
| Doctor consult | History, exam findings, diagnosis, plan | The clinical note | Here — the scribe drafts this |
| Prescription + advice | Glasses, drops, next review | Rx draft plus counselling | Draft from the same consult |
The map makes the boundary obvious. The scribe isn’t trying to replace your refraction record or read your OCT. It targets the one stop that turns a fast consult into a slow note, and leaves the rest of the flow alone. That’s the honest scope, and it’s why it works at volume: the busiest, most-repeated step is the one it takes off the doctor’s plate.
The volume math on a 90-patient session
No table can know your hospital, so here’s the arithmetic with every assumption showing. Time one session and swap in your own numbers.
| Line item | Written by hand | With the scribe |
|---|---|---|
| Consults per doctor per session | 90 | 90 |
| Writing time per note | ~1.5 min | drafted for you |
| Review-and-sign per note | included above | ~20 sec |
| Documentation time per session | ~135 min | ~30 min |
| Per month, 26 OPD days | ~58.5 hours | ~13 hours |
Ninety seconds per note is deliberately low, because eye notes are short. That’s the point: at OPD volume, even a short note times 90 is a couple of hours a doctor loses per session. Twenty seconds to review and sign is an assumption about you, not a product claim, so test it on your floor. Unlimited notes with no per-note metering is what makes this survive the arithmetic, because the cost doesn’t climb with the patient count.
”Aankh mein irritation”: Hindi in, English notes out
An eye consult in India is rarely one language. The patient reports the problem in Hindi or a regional tongue, the doctor answers in a mix, and the exchange is code-mixed: “aankh mein irritation, dhup mein zyada, do din se.” Single-language dictation breaks on that.
The scribe listens in English, Hindi, and 20-plus Indian languages, including code-mixed speech, and follows switches mid-sentence. What comes out is always a clean clinical English note, so the chart reads the same whether the consult ran in Hindi, Telugu, or both. The mechanics are in our post on Hindi and Indian-language capture.
What an eye hospital pays at OPD scale
We think flat pricing is the only model that survives eye-OPD volume. Per-note pricing punishes exactly the practice that documents the most: a high-throughput hospital running hundreds of encounters a day.
Patient Square is flat per doctor, ex-GST, billed with 18% GST added. On annual billing: Assist at ₹1,599/mo (₹1,887 with GST) covers the scribe, ICD-10 suggestions, prescription drafts, and scheduling. Copilot at ₹2,399/mo adds the bundled AI Copilot EHR and WhatsApp messaging. Autopilot at ₹3,999/mo adds the AI Receptionist and AI Follow-ups. Every plan: unlimited notes, no per-note metering, no setup fee. The full ladder is on the pricing page.
These per-doctor tiers are Practice Copilot, the fit for a solo or small-group eye clinic. A larger eye hospital running many chairs, an IPD, and hospital-wide records is a Hospital Copilot conversation instead: it’s a separate product, quoted on a demo rather than a public per-doctor tier. If that’s your scale,
Where the scribe stops, and when EMR software wins
Three limits, plainly.
It captures spoken findings, not device output. IOP, refraction, biometry, and OCT results enter the note only if the doctor says them aloud; the scribe won’t pull them from a slit lamp or an autorefractor. If machine integration is your core requirement, a scribe alone isn’t that product. For structured refraction records, IOL biometry, and imaging attachment, start with a dedicated system: our ophthalmology EMR buyer’s guide covers device integration and names when a full eye-clinic EMR is the better fit than a scribe.
It doesn’t make clinical calls. ICD-10 codes are suggestions, the prescription is a draft, and every note waits for the doctor’s review and signature. If you’re deciding whether to adopt a scribe or a record system first, our take on scribe-versus-EMR ordering walks the trade-off.
ABDM integration is on our roadmap, not live. If ABHA-linked records are a hard requirement this quarter, verify any vendor’s real certification status before signing. What ships today: notes born digital, encrypted in transit and at rest, audio processed in memory and never stored, handled to DPDP Act 2023 standards. Details are on the security page.
None of this needs trust in a blog post. Pick your heaviest OPD session, run the 7-day trial on it, and