For seventy-five years, Indian medicine has produced billions of consultations and learned almost nothing from them. Every diagnosis happens in isolation. Every prescription disappears into the pile. And every doctor — no matter how brilliant — is forced to practice without access to the one thing that would make them better: knowing what actually worked.
We've spent the last six months trying to fix that. Not by replacing doctors. Not by selling a chatbot. But by building the first piece of clinical software that closes the feedback loop — and gets smarter with every patient it touches.
What we mean by "closing the loop"
A typical OPD consultation in India lasts three to five minutes. The doctor takes a history, examines the patient, makes a diagnosis, writes a prescription, and the patient walks out. In ninety-five percent of cases, that's the last interaction. The doctor never finds out if the medication worked, if the patient came back worse, or if the diagnosis was wrong.
Multiply that by five billion OPD visits a year and you get a system that produces an enormous amount of clinical signal — and captures none of it. No feedback. No learning. No improvement.
How it works in practice
The mechanics are deliberately mundane. A patient books an appointment. The day before, they get a message in their language with a friendly conversational pre-screen — symptoms, history, concerns. By the time they walk in, the doctor has a structured summary on screen.
During the consult, the doctor dictates the prescription naturally. The system structures it, checks interactions, and delivers a clean digital Rx to the patient. Three days later, the patient gets an automated check-in. Six days, another. Twelve days, the outcome. Each response feeds back into the system. Each one teaches the differential engine. Each one teaches the doctor.
The day-3 escalations were the surprise. We thought we were building a tool for the doctor. Turns out we were also building an early-warning system for the patient.
Six months of actual numbers
- Follow-up compliance up 2.1× compared to the doctors' previous methods.
- 5–7 minutes saved per consult, primarily from pre-screening eliminating the history-gathering phase.
- 3 catches from differential suggestions — cases where the AI surfaced a condition the doctor later confirmed.
- 97% ASR accuracy on Hindi-English code-mixed clinical speech.
What we're building next
More languages (Tamil, Telugu, Bengali, Marathi are next), better differential ranking, and the start of Lisn Paws — our veterinary expansion entering closed beta in Q4 2026.
Anand Bodh is the founder of Lisn Health. Previously: Dharaksha (Shark Tank India). Currently raising seed.