Inspiration
Our inspiration came from a real-life problem we witnessed with our neighbor. For years, he struggled with a recurring health issue. His medical reports were filled with complex jargon he couldn't understand, and the advice he received was often conflicting. Because the early clues were missed, a manageable condition eventually escalated into a serious, life-threatening crisis. This made us ask a critical question: What if he could have just understood that first report? What if a simple, clear conversation could have empowered him to take the right action years ago? We built MediGuide AI to ensure no other family has to face that same preventable struggle. Our mission is to replace health anxiety with the confidence that comes from clear understanding.
What it does
MediGuide is your go-to health assistant. You can upload a medical report or even just a photo of a skin rash. Then, you just talk to it. Ask a question in your own language—we support 29 of them—and you'll get a simple, spoken answer back. It's designed to help with everything from understanding a scary diagnosis to planning a healthier diet for the week.
How we built it
We put this together with some amazing tools. We designed how it looks and feels in Figma. We built the actual app with Bolt AI, which let us build quickly. Everything that runs behind the scenes, like storing user data securely, is handled by Supabase. The brains of the operation is Google Gemini, which understands the reports and figures out the answers. The friendly voice you hear comes from ElevenLabs. And use of Expo for mobile app We have also used the VS code with Cline for few small edits.
Challenges we ran into
We were not able to revert back to the old update inside bolt, and we had to import the project from the github and start new One of our biggest technical challenges was creating a seamless, real-time conversational flow between our AI services. We had to architect a system where the text generated by Google Gemini could be instantly and smoothly streamed to ElevenLabs for voice output, creating a natural back-and-forth conversation. Another significant challenge was designing our Health Insights. It required building a dynamic system that could not just parse a single report, but synthesize data from various user uploads over time. Ensuring the AI could generate insights that were both personalized and context-aware was a complex task that demanded careful prompt engineering and data management. Implementing the Onboarding and family profile was also a chllange.
Accomplishments that we're proud of
Honestly, we're just really proud that it works the way we imagined. Seeing someone ask a question in their own language and get a clear, reassuring answer back is amazing. We're also proud that we built something that's useful for both scary medical emergencies and for simple, everyday health goals, all in one app.
What we learned
We learned the few limitation of vibe coding with bolt, securing databases with supabase is challenging and requires extra learning. We also learned how fast you can build a really good product with modern tools like Bolt AI and Supabase it let us focus on what really matters: the user.
Key MVP Features (How We Solve It):
- Easy Onboarding for Families Create an account and add up to 6 family members quickly. Why it matters: Sets up the app for family use from the start.
- Upload Any Report and chat with it (PDF or Image) Users can upload photos or PDFs of medical reports. OCR reads the text instantly. Why it matters: It’s the first step—makes complex info readable.
- Voice-Based AI Chat in 29 Languages Tap the mic, ask questions in your language, and hear simple answers. Why it matters: Solves confusion with clear, spoken responses.
- Critical Issue Alerts AI highlights serious issues like high BP right away. Why it matters: Adds urgency and real health value, not just translation.
- Ongoing Health Insights See a feed of health tips and trends from all your reports. Why it matters: Helps users track health, not just react.
- Diet & Exercise Tips from Reports Get food and activity suggestions based on your results. Why it matters: Offers real action steps, not just explanations.
What's next for MediGuide AI
We're just getting started—there are so many opportunities to improve.
First, we plan to make the AI chat even more powerful. We'll add chat tabs to keep conversations organized and give the AI a better memory so it understands your history. We'll also be exploring different AI models (LLMs) to find the absolute best one for health questions.
We plan to introduce a simple and fair token-based pricing model, so you only pay for what you use, or maybe per month pricinng.

Log in or sign up for Devpost to join the conversation.