Buddi

Our Story

We built Buddi around a simple idea: technology used in healthcare should feel human.

We were inspired by how easily health-related tracking can become focused on numbers, streaks, and performance. We wanted to explore a different approach—one centered on companionship, encouragement, and connection with a care team. That led us to create Buddi, a clinic-connected meal companion designed to support patients while giving clinicians useful, explainable information.

💡 The Idea

Buddi has two connected experiences.

For patients, Buddi is a mobile app with a friendly 3D companion. Patients can check in using their camera, leave optional notes for their care team, customize their companion, and spend time in a calming "Quiet Time" experience. The companion can become more energetic or settle into a gentle resting state based on engagement, but it is never designed to punish the patient or portray suffering.

For clinicians, Buddi provides a web dashboard where they can view patient check-ins, visit notes, care reminders, summaries, and observational alerts. AI can assist with meal-photo analysis and clinician summaries, but AI observes; clinicians decide.

A key principle throughout the project was making sure patients never see calories, macros, or numerical scores. We wanted the technology to support the person rather than turn their experience into a competition.

🛠️ How We Built It

We built Buddi as three connected components:

Patient app: React Native with Expo, featuring the 3D companion, check-ins, customization, milestones, and Quiet Time. Clinician dashboard: React + Vite interface for reviewing patients, alerts, summaries, notes, and care information. Backend: Express API with JWT authentication and a lightweight JSON data store. AI layer: Gemini as the preferred vision/summarization provider, with OpenAI as a fallback and deterministic mock AI available for demonstrations. The repository is organized into mobile/, dashboard/, and backend/, allowing us to develop the patient and clinician experiences independently while connecting them through the API.

📚 What We Learned

One of our biggest lessons was that building for a sensitive healthcare context requires thinking beyond whether a feature "works."

We learned to ask questions like:

Does this feature actually help the person using it? Could a seemingly positive game mechanic accidentally create pressure? What information should patients see—and what should remain clinician-facing? How can AI assist without replacing human judgment? How can we make a digital companion feel supportive without making it manipulative? We also learned a lot about connecting multiple applications through a shared backend, integrating AI into an existing product flow, and designing experiences where technical decisions and ethical considerations have to work together.

🚧 Challenges

One of our biggest challenges was balancing engagement with compassion.

Gamification can make an app more engaging, but traditional approaches such as scores, punishments, or competitive metrics didn't fit the experience we wanted to create. We therefore designed Buddi's companion to respond with calm resting states rather than negative consequences.

Another challenge was separating what AI can do from what a clinician should decide. We designed the system so AI can surface observations and generate summaries, while important care decisions remain with healthcare professionals.

We also had to bring together several moving pieces—a mobile application, a clinician dashboard, a backend API, 3D assets, authentication, and AI services—within a hackathon timeframe.

❤️ What We Built

Buddi ultimately became more than a meal-tracking interface. It became an experiment in how technology can make healthcare feel more supportive and less transactional.

Our goal wasn't to build an AI that replaces people. It was to build technology that helps people feel accompanied while giving their care teams better tools to support them.

Buddi: companionship over competition.