RxBridge

User Authorization
Dashboard
Viable Alternatives
Automated Emails

RxBridge: The AI-Powered Productivity Copilot for Pharmacists

Inspiration

Instead of building in a bubble, we started this hackathon by talking to people on the front lines of healthcare—a friend who has done extensive volunteering, an uncle who is an active researcher, and another friend on the front lines of the pharmaceutical industry. We asked them one question: "What are the most frustrating, time-wasting logistical chokepoints you see every day?"

A powerful theme emerged: the immense productivity drain caused by systemic supply chain failures. The most potent example was the sudden panic of a drug shortage. For a pharmacist, this isn't a news headline; it's a 20-minute, high-stress fire drill of manual research, phone calls, and paperwork that brings their entire workflow to a halt. We saw a critical productivity crisis and were inspired to build a tool that gives pharmacists back their most valuable asset: time for patient care.

What it does

RxBridge is an AI-powered dashboard that gives pharmacists a real-time, intelligent view of their supply chain to proactively manage and solve drug shortages & rapidly communicate with the prescribing doctor. For a busy pharmacist, it:

Proactively Detects Disruptions: Our backend agent autonomously scrapes public FDA databases for both drug shortages and critical medical device recalls, comparing this data against the pharmacy's own inventory.

Analyzes Real-World Impact: We didn't want a simple alarm. Our system uses the industry-standard "Days of Supply" metric to provide a tiered alert system (🔴 Critical, 🟡 Awareness, or 🔵 Routine Reorder), giving the pharmacist immediate, actionable insight into the true urgency of the situation.

Provides an Instant, Grounded Solution: For critical shortages, our Gemini-powered agent acts as an expert assistant. It first suggests medically valid therapeutic alternatives. Crucially, it then cross-references these suggestions with the pharmacy's own inventory to ensure the alternatives are actually in stock before presenting them as a viable solution.

Automates the Final Step: With a single click, the pharmacist can use the AI's validated suggestions to generate a concise, professionally formatted communication draft to send to the prescribing doctor's office, resolving the issue in seconds.

How we built it

We chose a modern, full-stack architecture to build a robust and scalable platform. Our entire design philosophy was grounded in reality: using genuinely available data, maintaining strict privacy by using zero PHI, and building a tool that could realistically exist.

Frontend: A dynamic and responsive dashboard built with Vite + React and styled with Tailwind CSS. It is hosted on Vercel.

Backend: A high-performance API server built with FastAPI (Python), which serves as the central orchestrator for our entire agentic workflow. It is hosted on Railway.

AI & Data Pipeline: We used BeautifulSoup for web scraping, a heavy dose of Pandas for data manipulation, and Rapidfuzz for robust, fuzzy text matching between the scraped data and our database. The core intelligence is powered by the Gemini API, which is carefully guardrailed with strict prompt engineering.

Database: A PostgreSQL database hosted on Supabase stores our pharmacy inventory and alert history.

A Note for the Judges: Our Alignment with Your Tracks

We were very intentional in designing RxBridge to be a direct and powerful solution for our chosen challenges:

For the Productivity Track: This project is the definition of helping a professional "work smarter, not harder." By automating the most chaotic part of a pharmacist's day, we directly reduce "cognitive overload" and make their workflow "feel manageable."

For the Commure Challenge: Our mission was to "simplify healthcare," and we did that by tackling a complex logistical workflow. Our "customer" is the pharmacist, and by making their day less stressful, we are ensuring "the day doesn’t end till the customer’s happy," which in turn leads to safer and more reliable care for patients.

For the Gemini API Challenge: We used Gemini for its most advanced capabilities. It acts as an expert assistant to "Analyze info" by suggesting therapeutic alternatives and to "generate creative content" by drafting professional communications, showcasing a sophisticated and practical use of the model.

Challenges we ran into

Our biggest challenge was ensuring the data pipeline was robust. Drug names from the FDA website and our inventory didn't always match perfectly ("Acetaminophen" vs. "Acetaminophen 500mg Tablet"). We solved this by implementing Rapidfuzz to create a smart, fuzzy matching system. We also dedicated significant time to prompt engineering, creating an "expert pharmacist" persona with strict safety boundaries to ensure Gemini's responses were consistently reliable and clinically appropriate.

Accomplishments that we're proud of

We are incredibly proud that we built a full, end-to-end agentic workflow that solves a real-world problem with zero reliance on protected health information. By focusing on public and operational data, we created a solution that is not only powerful but also ethically sound and realistically deployable. Our tiered alert system, based on the industry-standard "Days of Supply," is a feature that shows our deep empathy for the user's workflow.

What we learned

Our biggest takeaway was the power of a product-first, user-sympathetic development mindset. We didn't start with a cool piece of tech and look for a problem. We started by doing deep research into a real user's daily frustrations and then built a technology solution tailored to solve that specific pain point. This user-centric approach was our north star and is the reason our final product has so much practical value.

What's next for RxBridge

This is just the beginning. The vision is to make the alerts real-time and integrate directly with major Pharmacy Management Systems (PMS). We also plan to expand the agent's capabilities to monitor a broader supply.