How we measure Clinic Scribe’s accuracy
A percentage with no method behind it means nothing. Here are our numbers, how they were measured, and what still goes wrong.
Every number on this page comes from role-play consultations written by FrontivaAI and scored automatically. They are not real patients and not a clinical study. They tell you what we test and where the scribe is weak; your own pilot measures it on your patients before you rely on it.
At a glance
| Test (October 2026) | Key facts in the note | Serious errors |
|---|---|---|
| 60 full-length visits, typed scripts | 94.1% | 1, a contradictory instruction; signing was blocked by a check |
| 22 visits read aloud and recorded, then transcribed on our computer | 93.1% | 1, a relative’s medicine in the patient’s list; signing was blocked by a check |
| The same 22 visits as typed scripts | 97.4% | 0 |
| Recorded, Hindi (15 visits) | 95.4% | 1 |
| Recorded, English (7 visits) | 88.0% | 0 |
How the numbers are measured
Each test consultation comes with a list of the facts its note must contain - a dose, a medicine stopped, a symptom the patient denied, the follow-up - and the errors it must never contain, such as a stopped medicine written as continued. A program checks every note against both lists. The score is the share of key facts found; a serious error is any “must never” that appears.
- Written as real Indian OPD visits: mixed languages, relatives in the room, numbers said in English
- Recorded by the founder and friends reading the parts aloud - mostly one voice, in a quiet room
- Scored by fixed patterns, not by a clinician; a native-speaker and clinician review is still to come
- The test sets we tune on are kept apart from the ones we only measure on
What it still gets wrong
Most of what the recorded notes missed started with speech recognition, not with writing the note: the same scripts typed scored 97.4%. The weakest point is drug names, as it is for every speech system - a brand can come out as a near neighbour or as ordinary words. Each of these is flagged for the doctor rather than corrected silently.
- Drug names misheard (“Donapasil” for donepezil) - flagged with the likely name
- English recordings scored lower than Hindi in our tests
- A number heard wrongly (“pachchees” as “pachaas”, 25 as 50)
- One person reading two parts sounds like one speaker - real clinics have two voices
Why we publish this
An accuracy figure without its method cannot be checked. These numbers are re-measured whenever the scribe changes, and in your pilot we agree the success measure in writing before day one and measure it on your own consultations.
See a sample note · Book a demo
Frequently asked questions
Does the audio go to the cloud?
No. It is transcribed and the note is written on a computer installed in your clinic. The audio is deleted once it is transcribed, and every deletion is logged.
Does it diagnose or prescribe?
No. It writes up what was said. You review, correct and sign every note; nothing is filed without your signature.
What if a patient does not want to be recorded?
Tap “Patient declined” and nothing is recorded. You can dictate a short summary yourself after the patient leaves, and the note is marked as dictated.
Can FrontivaAI see our notes?
We do not open them. Notes are kept on the computer in your clinic, and the app opens only with your clinic’s passphrase. Our private, encrypted link to that computer exists to install updates and restart the app, not to read notes.
What happens if the power or the WiFi goes?
The clinic computer runs on a UPS, so a short power cut changes nothing, and in a long one it shuts down cleanly and starts again when power returns. If the WiFi drops, a recording waits on the tablet until the clinic computer confirms it has it. The internet is never needed.
What happens to our notes if we stop?
They are yours. Signed notes can be printed or saved as PDFs at any time, and the nightly encrypted backup is on a drive in your clinic. Before the computer leaves, we hand over that backup and its key.
How do we stop the service?
The two-week pilot comes first, with no obligation if the agreed success measure is not met. The service agreement states the minimum term and the notice period in plain words before you sign; ask us for it during the demo.
How is it updated?
We install updates over a private encrypted link, outside clinic hours. Nobody at the clinic has to do anything, and an update never touches your notes.
How Clinic Scribe works, or book a demo in your clinic.
₹15,000 a month per doctor plus GST: about ₹580 a working day, or under ₹20 a patient at 30 patients a day. The computer, microphone and tablet are included, and the first two weeks are free.
All Clinic Scribe pages
Start here
Check it yourself
By specialty
- Psychiatry
- IVF & fertility
- Dermatology
- Diabetes & endocrine
- Paediatrics
- Neurology
- Pulmonology
- Urology
- Dietitians
- General physicians
Read more
If this sounds familiar
- We keep missing calls - what is that costing?
- Customers ring after we close
- The line is engaged all morning
- Should I hire a receptionist?
- People book and then do not turn up
Home · Pricing · Use cases · Industries · Comparisons · Products · Blog · Book a demo · All solutions · Hear a demo call
For doctors: Clinic Scribe, an on-premise AI medical scribe
Legal: Privacy Policy · Terms of Service · Cookies · Image credits · Security and trust · Acceptable use
{/* __FV_PRIVACY_V1__ */} {/* __FV_FOOTLINKS_V3__ /solutions/ had ZERO inbound links and it is the page that lists every comparePages entry, including all 25 service x city pages. */}AI receptionist for dental clinics · salons · coaching institutes · real estate agents · hospitals · diagnostic labs · AC repair · plumbers · electricians · appliance repair
Common questions: what it costs in India · does it speak Hindi · does it speak Telugu · does it speak Tamil · does it speak Kannada · does it speak Malayalam · does it speak Bengali · does it work with call forwarding · does it send a WhatsApp confirmation · can it transfer to a person · what if it does not understand
Compare: vs voicemail · vs a chatbot · vs forwarding to your mobile · Vapi alternative · Retell AI alternative · Bolna alternative · Exotel alternative · Knowlarity alternative · Ozonetel alternative · Gnani alternative · Yellow.ai alternative · SquadStack alternative · Haptik alternative · virtual receptionist India
{/* __FV_FOOTLINKS_V2__ these and the four new language pages were in Navbar.tsx only, which is React and never reaches a crawler - the same miss as __FV_CITY_CRAWL_V1__ */}AI receptionist by city: across India · Hyderabad · Bengaluru · Chennai · Delhi NCR · Mumbai · Pune · Kolkata · Ahmedabad · Jaipur · Surat · Lucknow · Indore · Bhopal · Nagpur · Chandigarh · Visakhapatnam · Vijayawada · Coimbatore · Madurai · Kochi · Mysuru