A Doctor in Myanmar Asked Why We Still Write Notes by Hand. 12 Years Later, He Built the Answer
It started in 2012, in a medical ward at a general hospital in Myanmar.
It started in 2012, in a medical ward at a general hospital in Myanmar.
I was a junior doctor, sitting in the ward after a long shift, writing discharge notes by hand. Not one or two. Sometimes a dozen. Each one a full summary — the patient's condition, every treatment given, every lab result, every procedure — all written out manually, page after page, while patients waited outside.
There was a computer in the room. One. Shared by everyone. Mostly used for PowerPoint presentations at seminars. The internet was so slow it was barely usable. Mobile phones didn't even have internet yet.
And I sat there thinking: why are we still doing this?
That question never left me.
A Career Built Around One Problem
I graduated from the University of Medicine in 2013. But my career didn't follow the typical clinical path. It wandered — deliberately — through public health, then into IT, then deeper into healthcare informatics. I completed my Master's degree in Healthcare Informatics at the University of Wisconsin-Milwaukee. Along the way I took courses in information systems and design, picking up knowledge wherever I could find it. I worked with UNOPS during COVID-19, managing health data at scale during one of the most chaotic periods in modern healthcare history.
Every role, every field, every experience taught me something different. But no matter where I worked or what I was doing, the problem I saw in that ward in 2012 stayed the same.
Clinics in Myanmar wanted to use technology. They talked about it. They dreamed about it. But the real options were almost nonexistent. Enterprise systems like Epic were built for large, well-funded hospitals in wealthy countries — expensive, complex, and completely disconnected from the reality of a small clinic in Yangon. Open-source platforms like OpenMRS were technically available but so complex to configure that they created more problems than they solved. And budget was always the final wall — in a healthcare system where the majority of resources go toward medicine and keeping patients alive, information systems feel like a luxury nobody can afford.
So I did what I could with what was available. And I kept doing it for nearly a decade.
It wasn't just clinic systems. Over the years I built a medical dictionary app, a drug encyclopedia, an OPD wait time search tool, a COVID-19 data collection system with a real-time dashboard, health information websites, and more. Some were small. Some were used by thousands. Most people who used them never knew who built them.
At the time, each project felt standalone. A problem someone had, a tool I built to solve it. But looking back now, I can see what was actually happening.
Every single thing I built was a piece of a puzzle.
The medical dictionary taught me about clinical terminology standardization. The COVID dashboard taught me about data pipelines and real-time reporting at scale. The drug encyclopedia taught me about how doctors actually search for and consume clinical information. The clinic systems taught me about workflow, privacy, access control, and the gap between what a system can do and what a doctor will actually use under pressure.
None of it felt like preparation at the time. But all of it was. By the time I started building Medaius, I wasn't guessing about what clinics need. I had spent a decade learning it the hard way — by building, failing, iterating, and listening.
A Decade of Building — and Hitting Walls
From around 2017, I started building systems for clinics using whatever tools I had access to. It started with Excel and Microsoft Access. Then Google Sheets. Then AppSheet — which was a genuine step forward, flexible enough to build real workflows, accessible enough for clinics with limited technical capacity.
The clinics I helped had different needs. A specialty ward inside a hospital that wanted to track patient outcomes for research. An online clinic that was growing fast and needed proper record management. A mental health clinic with unique session workflows that nothing off the shelf could handle.
Each one was different. Each one taught me something. Sometimes I was compensated for the work, sometimes I contributed my time — but the motivation was always the same. I couldn't watch colleagues struggle with paper and spreadsheets when I had the skills to help.
And those systems worked. The online clinic platform I designed eventually held over 100,000 patient records and handled millions of visits. Doctors who didn't even know who built the system were using it every day. That meant something to me.
But I always knew these tools had a ceiling. AppSheet is powerful for what it is — until your data grows, your workflows get complex, and your clinic starts asking for things the platform was never designed to handle. The system would start straining. Features would hit limits. Customization would become painful.
And every system I built eventually hit that wall.
I'd patch it, extend it, find workarounds. But the wall was always there. And over time, I started to understand that the problem wasn't the tools. The problem was that no one had built the right foundation.
The Version That Taught Me Everything by Failing
In 2023, I decided to stop working around the limitations and build a proper system from scratch. Python, Django, pure HTML and CSS. No frameworks, no shortcuts — just code.
I spent more than half a year on it. It had all the core features a real system needed — patient management, encounter notes, basic team management, user roles. I was proud of it. For a while, it felt like the thing I had always wanted to build.
But then the problems started showing up.
It wasn't mobile responsive. Adding a new feature meant touching dozens of tightly coupled HTML files. The UI was rigid in ways that would make customization for different clinic types nearly impossible. I could feel the ceiling getting closer with every commit.
And slowly, I had to admit it to myself: I was hitting the wall again. Just at a higher level.
In late 2024, I suspended the project.
That decision was harder than it sounds. More than half a year of work, shelved. But I had learned things I couldn't have learned any other way — about architecture, about what a clinic system actually needs to be flexible, about where my own technical limits were. The failure wasn't wasted. It was research.
This time, I knew exactly what I needed to build differently. I just needed to find the courage to start again.
How AI Changed What Was Possible
Something had been quietly shifting since 2023 — the way I could work with AI as a development partner.
Before that, fixing a single coding problem could cost me hours. I was strongest in Python but uncertain in most other languages. JavaScript frameworks felt foreign. The gap between what I could imagine and what I could actually build was wide and frustrating.
AI didn't close that gap magically. Anyone who tells you AI just writes your code and everything works is lying. It hallucinated. It introduced bugs. It sometimes created more mess than it solved. Every single suggestion needed supervision, testing, and often complete reworking.
But what it gave me was something more valuable than code. It gave me a thinking partner. When I needed to evaluate three different architectural approaches for handling multi-clinic data isolation, I could talk it through. When I wasn't sure how to design an auditable AI approval workflow for clinical records, I could explore options. When I was stuck at 2am with a bug I couldn't find, I had somewhere to turn.
My productivity didn't double. It multiplied.
In early 2024, I started over. Next.js for the frontend. Python for the backend — deliberately separated for scalability and security. Redis for caching. Proper DevOps. More than 500 commits, each one an iteration. I tested every feature not just as a developer but as a doctor — asking myself at every screen: if I were seeing patients right now, would this get out of my way or add to my burden?
I wore every hat at once. Backend engineer. Frontend engineer. UX designer. Doctor. Clinic administrator. Security reviewer.
Ten months later, Medaius was production-ready.
What Medaius Actually Is
After everything — the ward in 2012, the decade of building, the Django version that hit its wall, the ten months of iteration — what did I end up with?
Not an EHR. I want to be clear about that. EHR is a word that means something very specific in healthcare IT, and it carries a lot of baggage. Complex. Expensive. Built for compliance, not for doctors. Built for billing departments, not for the person sitting across from a patient.
Medaius is a clinic workspace. Built with an operating system philosophy — meaning it's modular, it's flexible, and it's designed to grow with the clinic rather than force the clinic to grow around it. A general practice and a mental health clinic and a dental practice all have fundamentally different workflows. Medaius handles all of them, because clinic managers can configure it to match how they actually work rather than how a software company thinks they should work.
The AI features are built in from the ground up — not added on top as an afterthought. Ambient transcription during consultations. Automatic summarization of patient histories. Extraction of data from PDFs, scanned documents, even handwritten notes. And crucially — every single AI output requires a doctor to review and approve it before it becomes part of the official record. Because AI should assist clinical judgment, not replace it. That's a principle I won't compromise on.
It's subscription based, with modular pricing so clinics only pay for what they actually use. There's a sponsor module for NGOs and insurance companies. A reporting module for aggregate data. And if a clinic isn't ready for AI at all — that's fine. They can disable it entirely and use everything else.
Right now, nearly 300 doctors are ready to start using it.
That number still surprises me sometimes. But then I remember — they've been waiting for something like this for a long time. We all have.
Why I Built This
I no longer practice as a clinical doctor. People sometimes ask why I left.
The honest answer is: this is what I was supposed to do.
No one else has the exact combination of experience I have — clinical medicine, public health, healthcare informatics, AI, and a decade of building tools for the exact clinics Medaius is designed for. I know these workflows from the inside. I know the constraints. I know what doctors actually need versus what they think they need.
If I don't build this, no one will build it quite like this.
So I built it.
This Wave Won't Wait
Healthcare technology is changing faster than at any point in my career. AI is no longer a future concept — it's being built into clinical workflows right now, today, by teams around the world racing to define what modern healthcare software looks like.
I watch this race closely. And what worries me is not the competition. What worries me is the clinics that get left behind.
The big players will build AI tools for big hospitals. They always do. But the small clinic in Yangon, the online mental health practice, the specialist running a two-room clinic in a mid-size city — they deserve this technology too. They need it more, because they have fewer resources to waste on inefficiency.
That's the wave I'm trying to catch. Not for the technology itself, but for what it means for the doctors and patients who have been ignored by every previous wave of healthcare innovation.
We can't afford to miss this one.
Medaius is live at medaius.com. If you're a clinic looking for a better system, a doctor who recognizes this problem, or someone who wants to help bring this to the clinics that need it most — I'd love to hear from you.
We're just getting started.