MOMent

What inspired us

We were inspired by a simple problem: generic AI doesn't always understand the context behind a health question. In pregnancy, a symptom that seems normal on its own can mean something very different depending on the mother's week of pregnancy, medical history, medications, or previous reports. We wanted to build something that could remember that context instead of treating every question like a completely new conversation.

What we learned

While building MOMENT, we realized that healthcare AI isn't just about getting better answers. Safety has to come first. We learned that some decisions, especially emergency triage and medication-related questions, should not depend entirely on an AI model. This led us to combine AI with deterministic safety rules and trusted medical sources.

How we built it

We built MOMENT as a full-stack application using React, TypeScript, FastAPI, and Python. It combines longitudinal pregnancy memory, a clinical safety engine, evidence-based RAG, Gemini, OCR for medical reports, medication tracking, food-safety checks, emergency hospital discovery, and a "Since Your Last Visit" doctor summary.

One important design choice was keeping a human verification step when extracting information from medical reports, rather than automatically saving OCR results into the patient's history.

Challenges we faced

Our biggest challenge was figuring out where AI should help and where it should step back. We had to connect multiple systems while keeping the experience simple and safe for a non-technical user.

Building MOMENT taught us that a meaningful healthcare product isn't about adding AI everywhere. It's about using AI where it helps, adding safeguards where it matters, and knowing when a human healthcare professional needs to take over.