MedScribe

Inspiration

MedScribe came from a problem we've all lived. We've walked out of appointments and immediately forgotten what the doctor said. We've stared at a prescription with no idea where to fill it. We've Googled symptoms at midnight wishing someone had just told us what to do at home.We didn't find this problem in a textbook ,we found it in our own waiting rooms and family group chats. MedScribe is the tool we wished existed the moment we left the doctor's office.

What it does

MedScribe is a patient portal that turns doctor visits into clear, actionable information. After an appointment, patients get a plain-language summary of their visit — no medical jargon, just what happened, what it means, and what to do next. Voice-recorded visits include audio playback so you can revisit the conversation anytime. When you're prescribed medication, MedScribe shows nearby pharmacies where you can pick it up — with hours, distance, and directions. But we go a step further: for every diagnosis, we recommend relevant home remedies alongside your prescription, bridging the gap between clinical care and everyday wellness. Patients also complete intake forms digitally before their visit — medical history, conditions, allergies, and documents — so they never have to fill out the same clipboard twice.

How we built it

Our backend runs on Flask, handling routing, data processing, and serving the application. For the voice-to-text feature, we integrated the Groq API — when a doctor records a visit, the audio is transcribed and processed into structured, patient-friendly summaries. The frontend is built with pure HTML, CSS, and JavaScript. We designed a glassmorphism UI with animated backgrounds, frosted translucent panels, and responsive layouts that work seamlessly on both desktop and mobile — no frameworks, just clean code. Throughout development, we leveraged AI tools heavily. Cursor was our primary coding environment. Claude, ChatGPT, and Google Gemini each played a role in brainstorming features, debugging, generating UI components, and refining our design system. Using multiple AI tools let us cross-reference outputs and iterate faster than any of us could have individually.

Challenges we ran into

The biggest hurdle was cost. Many of the APIs we initially planned to use weren't free, which forced us to pivot mid-build — swapping services, rewriting integration code, and finding alternatives that fit within a hackathon budget. Every switch meant re-testing endpoints and adjusting how our backend handled responses. Getting the voice-to-text feature working was its own battle. Translating raw audio into clean, structured patient summaries isn't as simple as plugging in an API. We dealt with transcription accuracy issues, formatting inconsistencies, and the challenge of turning messy spoken dialogue into something that actually reads well as a medical summary. It took multiple iterations to get it to a place we were proud of.

Accomplishments that we're proud of

The moment our app took a clinical doctor's report and turned it into something a patient could actually understand — that was the breakthrough. Medical jargon like "recurring occipital cephalalgia" becomes "headaches in the back of your head." It sounds simple, but getting that translation right while keeping it medically accurate was a real challenge we're proud to have solved. We're also proud of the prescription finder. Patients don't just see what they're prescribed — they see exactly where to go get it, how far it is, and when it's open. It's a small feature that solves a surprisingly overlooked problem in healthcare apps.

What we learned

We learned that the hardest part of building a health website isn't the tech — it's the translation. Converting medical language into something accessible without losing accuracy forced us to think like patients, not developers. We also learned to be resourceful when tools don't go your way. When APIs we planned on weren't free, we didn't stall — we adapted, found alternatives, and kept building. Working with multiple AI tools taught us that no single tool has all the answers, but combining them makes you significantly faster and more creative. Most importantly, we learned that the best products come from solving problems you've actually experienced yourself.

What's next for MedScribe

MedScribe is just getting started. Here's where we're headed: Doctor-side integration is our top priority. Right now MedScribe serves patients — but we want to build the other half. A dedicated doctor dashboard where physicians can record visits, review auto-generated reports, manage patient records, and push summaries directly to their patients' portals. One appointment, one recording, both sides served instantly. Prescription price comparison is the natural next step for our pharmacy finder. We plan to integrate real-time pricing data so patients don't just see where to fill their prescription — they see what it costs at each location and which pharmacy is the cheapest nearby. Nobody should overpay for the same generic medication just because they went to the closest store. Privacy and security are non-negotiable as we scale. We plan to implement end-to-end encryption for all medical data, HIPAA-compliant storage, and role-based access controls so patients own their information and decide exactly who sees it. We're also exploring on-device voice processing so audio never has to leave the patient's phone. Beyond that, we're looking at personalized home remedy recommendations powered by verified wellness databases, multilingual support so language barriers never prevent someone from understanding their health, real-time medication reminders, and in-app messaging between patients and their care team. We built MedScribe because the patient experience is broken. The next chapter is making it complete — for every side of the conversation.