Overview

Indian doctors split their day between hospitals, wards and OPD queues, and their patients' records are split across three to five systems. Chikitsa, built by VigorusAI, brings it all into one hospital management system: reception and insurance claims, the nursing station, the IPD doctor, billing, pharmacy and analytics on the web, and a doctors' app that carries the OPD in a pocket. I led the research and designed the product end to end, from 40 hours of shadowing in three hospitals to the web system and a 39-screen doctors' app.

Details

Role: Product designer, research to final UI
Platforms: Android app for doctors, web for hospital staff, TV token board for waiting areas
Research: 40+ hours of shadowing, 12 interviews, a 2-week diary study, 8 products benchmarked

The doctors' app and the web system, built on one patient record. All patient data is placeholder.

The problem

Five systems, one patient, and a doctor who is always running late.

I spent more than 40 hours shadowing doctors, nurses and front-desk staff in three hospitals in Jaipur and Delhi. The same story repeated on every floor: patient data scattered across three to five systems, reports buried in long records, and staff who had already been burned by software too complex to learn. Doctors were spending around 40% of each appointment on administration. Affinity mapping 180+ observations gave us five themes.

Two people the system had to work for:

Dr Meera runs the consultation; Mr Ravi runs the desk. Both were losing hours to disconnected systems.

Research

Watching, not just asking.

Ethnography: 40+ hours shadowing across 3 hospitals in Jaipur and Delhi.
Interviews: 12 in-depth with 6 doctors, 3 admin managers, 2 nurses and 1 IT admin.
Diary study: 4 doctors logged their workdays for 2 weeks.
Benchmarking: 8 products, including Suki, Augnito, MyScribe, Simbo, Abridge, HealthPlix, Eka Care and Driefcase.
Synthesis: 180+ observations, affinity mapped into five themes.

Competitive Benchmarking

Voice and AI documentation tools from India, the US and Germany, compared on overview, features and language support.

Doctors had normalised the small inefficiencies, so they never mentioned them in interviews. They only showed up when we watched: prescriptions sent on WhatsApp, notes finished after the last patient, a report hunted across three tabs.

One record, every role

One patient record, six roles around it.

The site map grew to six areas, but the design rule stayed simple: every role works on the same patient record, from its own side. Reception registers and admits, the insurance desk files pre-authorisation and claims, nurses record vitals and care, doctors visit and prescribe, the cash desk bills, and administrators see the whole hospital. Nobody re-types what someone else already entered.

Assign Bed to Patient:

Reception assigns beds floor by floor; a bed shows who is in it with one hover.

The doctors' app

The OPD, in a doctor's pocket.

Many doctors consult at more than one hospital in a day and see OPD patients between ward rounds. A desktop EMR waits for them at one desk. So the doctor's side of Chikitsa became an Android app: 39 screens covering sign-in, the OPD queue, vitals, a five-step prescription with voice and AI, patient history, shifting patients and appointments.

Decision 1

Sign in the way Indian doctors already identify.
Mobile OTP, Healthcare Professional Registry (HPR) ID or email. Then the doctor picks the hospital they are at today, and can switch from the dashboard.

Decision 2

The queue at a glance.
Accepted, pending and cancelled appointments on the dashboard. The OPD list expands in place to accept a patient, and missed and closed patients sit in their own tabs.

Decision 3

A prescription in five thumb-sized steps.
History or voice, symptoms with severity, investigations, medicines, then advice and follow-up: one decision per screen, Next always under the thumb, and Generate with AI to draft from the conversation. Usability testing cut prescribing from 4 clicks to 2.

Decision 4

The doctor's own shortcuts.
Frequent medicines, complaints and advice live in personal masters and come back as one-tap chips.

Decision 5

Hand-offs without a phone call.
Shift a patient to another doctor or admit them to IPD from the patient screen, with a reason and the urgency.

The web system

Behind the counter: the web system.

On the web, the same record serves the hospital's desks: bed assignment by floor and ward, the IPD patient record with vitals history, insurance pre-authorization through to settlement, billing and pharmacy, a token board for waiting rooms, and an overview of the whole hospital for administrators.

The IPD patient record: vitals, trends and history on one page.

Process and testing

Tested with the people who would use it.

Three workshops produced 60+ concepts, and an impact versus effort matrix picked 15 for the first release. Over six weeks the screens went from paper sketches of 12 key screens, to digital wireframes with cognitive walkthroughs, to high fidelity. Then moderated think-aloud sessions with 5 doctors and 3 admin staff: 60 minutes and 8 tasks each. Three findings changed the design.

Every critical finding was fixed and re-tested before development.

Impact

40% less time on administration.

Operations: 40% less administrative time, 60% faster record retrieval, 75% fewer scheduling conflicts, 35% fewer billing errors.
Adoption: 89% user satisfaction, 94% daily use within 2 weeks, 67% fewer support tickets.
Clinical: 25% more patients seen per day, 15% better prescription accuracy, 82% of doctors reported better decisions.
Method: Compared with the baseline from the diary study and shadowing, measured over the first 4 weeks at the launch hospitals.

This is the first EMR that actually helps me practice medicine.

~ A Doctor

Training new staff takes days instead of weeks.

~ A front-desk manager

Reflection

Watch first, then ask.

Watching people work showed the inefficiencies they had stopped noticing. Card sorting showed that doctors and administrators organize information differently, so the IA follows each of their mental models instead of one compromise. And AI earned trust only when it showed its reasoning and left the final call to the doctor. Three rounds of testing caught all of this before a line of code was written.