Inspiration During our second year of college, we found ourselves drowning in deadlines, sleepless nights, poor eating habits, and emotional exhaustion. One teammate even silently battled anxiety, while another was skipping meals due to lack of time and awareness. We realized that wellness wasn't just about going to the gym or eating salad—it’s deeply personal, messy, and often neglected by students like us.
We imagined: What if we had something that not only understood our state of mind, but also acted to improve it? That's how WellNest was born—our AI-powered wellness companion, designed to care for students the way we wish we had been cared for.
Inspired by our own struggles, we set out to build a tool that can guide, support, and even take small steps on behalf of students to keep their physical, emotional, and mental health in check—one gentle nudge at a time.
What it does WellNest is an AI-powered student wellness companion that supports your health journey through intelligent, personalized care. It:
Tracks your mood using text, voice, or selfies, and suggests music therapy to reduce stress or boost focus.
Creates personalized workout videos (like yoga, stretching, or facial exercises) based on your time and energy level.
Designs a custom diet plan based on your wellness goals—like glowing skin, building muscle, or managing health conditions.
Uses Agentic AI to take real action for you: booking virtual therapy or fitness classes, ordering supplements or groceries, and recommending mental health check-ins.
It’s not just an app—it’s your smart wellness buddy that understands your needs and acts on your behalf to make wellness easier, smarter, and more student-friendly.
How we built it Frontend:
Developed a cross-platform mobile app using Flutter for a seamless experience on both Android and iOS.
Integrated simple, user-friendly UI/UX to make wellness tracking quick and intuitive.
AI & Agentic Systems: Used OpenAI’s GPT-4o to generate personalized diet plans, exercise routines, and weekly wellness summaries.
For mood detection, we used:
Sentiment analysis on user input (text, emoji, or journal logs).
Optional facial emotion recognition using MediaPipe (for camera-based detection).
Implemented an agentic AI workflow inspired by LangGraph and ADK to:
Auto-book therapy sessions and fitness classes.
Place grocery/supplement orders based on meal plans.
Suggest community wellness events based on user mood and needs.
APIs & Integrations: Spotify API for mood-based music therapy suggestions.
Zoom / Google Meet API to auto-book wellness webinars or counseling sessions.
Grocery APIs like BigBasket or Amazon Fresh to simulate real-time ordering.
Firebase for user authentication, data storage, and cloud messaging.
Wellness Dashboard: Built a personalized dashboard using Chart.js for tracking user progress (mood trends, workouts completed, meals followed, etc.).
Challenges we ran into Integrating Agentic AI Behavior:
Designing workflows where the AI could act on behalf of the user (e.g., booking sessions or ordering groceries) was complex. We had to simulate real-world actions and chain multiple decisions based on user input, which required careful logic planning using multi-agent frameworks.
Balancing Personalization and Simplicity: It was challenging to make wellness recommendations highly personalized without overwhelming the user. We had to fine-tune prompt engineering to keep responses helpful, short, and friendly—especially for students new to wellness tracking.
Mood Detection Accuracy: Accurately detecting mood from text, voice, or facial expressions while ensuring privacy was difficult. We had to experiment with lightweight models and fallback options like emoji-based mood input.
Time Constraints: Building a full-stack AI app with multiple wellness modules (diet, fitness, mood, agents) in limited hackathon time was a big stretch. Prioritizing features and optimizing development speed was a constant decision-making challenge.
API Limitations & Fake Integrations: Some APIs (like Spotify or grocery services) had limited access or required paid plans, so we had to simulate parts of the experience to show functionality without real-world purchases.
Accomplishments that we're proud of Built a Functional Multi-Feature Wellness App:
We successfully developed an AI-powered app that integrates diet planning, custom workouts, mood tracking, and music therapy—all tailored for students.
Implemented Agentic AI Workflows: We’re proud that our app doesn’t just suggest—it acts. It can automatically book wellness sessions, order health items, and guide students without manual effort.
Seamless Mood-to-Music Integration: We built a working system where mood detection is linked to personalized music therapy—helping students manage stress in real-time.
Created a Personalized Wellness Experience: From mental health to physical goals, our app adapts to each user’s journey. Whether it’s glowing skin or anxiety relief, students get relevant care at their fingertips.
Built All of This in Just a Few Days: Despite the limited hackathon timeline, we brought together AI, mobile development, UX, and agent automation into a fully working prototype—something we didn’t think was possible at the start!
What we learned Understanding Agentic AI in Real Life:
We deepened our knowledge of Agentic AI—how to go beyond chat-based responses and build AI that can make decisions and take real-world actions to support users.
Prompt Engineering Matters: Crafting the right prompts for GPT and other AI tools was crucial. We learned how small changes in language can drastically improve personalization and user experience.
Designing for Student Wellness Is Complex: Wellness isn’t one-size-fits-all. We learned how to balance multiple dimensions—mental, physical, emotional—and create a system that feels human, empathetic, and supportive.
🧩 Integrating Multiple APIs Smoothly: Working with tools like Spotify, Firebase, and video call APIs taught us how to manage real-time integrations, handle authentication, and simulate agent actions.
Time Management and Team Collaboration: Coordinating different features under a tight deadline helped us improve task prioritization, divide work efficiently, and support each other throughout the hackathon.
What's next for WellNest Launch a Public Beta for Students:
We plan to refine the UI and release a beta version for college students to test, give feedback, and help us shape a real-world version of the app.
🤝 Partner with Universities & Wellness Experts: We're exploring collaborations with campus counselors, dietitians, and fitness coaches to bring verified wellness plans and real-time support into the app.
🧠 Improve Agent Autonomy with Real Actions: We'll expand the agentic capabilities to handle tasks like setting reminders, checking class schedules before suggesting wellness activities, and syncing with calendars for better personalization.
🧬 Add AI-Driven Health Insights: We want to introduce weekly wellness reports powered by AI, helping students visualize mood trends, habits, and health improvements in a meaningful way.
🔐 Focus on Privacy & Ethics: As we grow, we’ll prioritize building robust data privacy protections and ethical AI practices, especially since we’re working with sensitive mental and physical health data.
🌍 Scale Beyond Students: While WellNest was born for students, we envision expanding it into a broader wellness tool for young professionals and communities seeking mindful, personalized support.
Log in or sign up for Devpost to join the conversation.