Chronicare

We’re building Chronicare, an app for IBD patients to track their condition, spot patterns, and connect with peers. For the hackathon, we built a voice-first agent that turns symptom tracking into a fast, empathetic conversation. Patients simply speak; the agent interprets, structures, and logs their health data in real time.
This replaces a painful reality: most patients still rely on pen and paper, improvised spreadsheets, or scattered notes. Our agent removes the friction that causes poor adherence and incomplete reporting.
Working prototype stability: The prototype successfully fetches the users symptoms and overall health based on the interaction with the voice agent. At the end of the call, the data gets stored in the backend
Technical complexity: The agent turns live speech into structured medical data by coordinating ASR, LLM reasoning, intent detection, and backend tool calls. It must reliably extract symptoms, severity, and wellbeing from natural language and map them to controlled formats. Behind a simple conversation, the system manages real-time audio processing, schema enforcement, and secure data storage to produce clinically meaningful records.
Real-world impact: IBD patients are expected to monitor symptoms, mental wellbeing, bowel habits, medication effects, and diet changes over time. This data is essential for clinicians to optimize treatment, yet current tracking methods are tedious and unreliable, leading to gaps, poor recall, and delayed interventions. A voice-native agent makes consistent tracking dramatically easier and more human. Better adherence means richer longitudinal data, earlier detection of worsening disease, more precise clinical decision-making, and ultimately improved patient outcomes.
Theme alignment: The project demonstrates how AI and voice can be used to make a very important, but tedious task easier and more human. Speech is captured, interpreted by the model, transformed into structured medical data, and instantly reflected in the app with zero manual input.

As a startup, we already have an established app running on the App Store, with the tracking modules implemented. So what was built in this hackathon was this feature where the users could speak to this agent.