VeriWell AI Wellness Coach

Inspiration

Our core motivation stemmed from observing the growing mental and physical well-being crisis among college students. This demographic often navigates immense academic pressure, leading to chronic stress, burnout, and unsustainable health habits. The existing wellness tools are largely passive and fail to provide contextual, proactive guidance.

We set out to develop a solution that doesn’t just track metrics but understands the human element—a proactive, intelligent companion, VeriWell AI, capable of transforming raw data into truly actionable steps. Our ultimate vision is to create a scalable health foundation that can benefit anyone seeking deeper control over their long-term well-being.

What it does

VeriWell AI functions as a highly personalized, intelligent health and wellness system built on a foundation of user-generated data. Users securely log critical daily and temporal metrics, including:

Quantitative Metrics: Sleep duration, exercise time, and physical activity levels.
Subjective Metrics: Daily perceived stress levels and mood scores.

The platform then immediately leverages the Google Gemini API as its core analytical engine. The AI analyzes these habits, identifies correlation patterns (e.g., how exercise time impacts sleep quality and stress scores), and delivers personalized, actionable insights and evidence-based recommendations. This shifts the user experience from simple tracking to contextual, AI-driven improvement.

Furthermore, users can engage with Veri AI, a conversational interface powered by Gemini, which provides instant, empathetic advice and explains complex health trends in an intuitive and accessible manner. All personal data is managed and secured using the real-time capabilities of Firebase.

How we built it

VeriWell AI was engineered as a robust, full-stack application centered around three powerful, integrated technologies:

Frontend (Streamlit): We utilized Streamlit for rapid prototyping and deployment, allowing us to build a highly interactive, data-visualization-rich, and mobile-responsive web interface quickly.
Data Backend (Firebase): Firebase (specifically Firestore) serves as our secure, scalable NoSQL data backbone. It handles user authentication, stores all key metrics in real-time, and enables low-latency data retrieval necessary for immediate AI analysis.
AI Intelligence (Google Gemini API): The Google Gemini API provides the core intelligence layer. We engineered specific prompt frameworks to enable Gemini to perform complex tasks:

Trend Analysis: Detecting longitudinal patterns in user data.
Contextual Recommendation: Generating personalized, safe, and effective advice based on identified patterns.
Conversational Logic: Powering the Veri AI chatbot for intuitive user interaction.

This architecture ensures a smooth, end-to-end user experience, where data logging immediately translates into AI-powered insight.

Challenges we ran into

Integrating cutting-edge AI and database technologies presented several rewarding technical hurdles:

Google Gemini API Integration & Prompt Engineering: We faced challenges in maintaining fault tolerance and consistent latency when interacting with the external Gemini API. We extensively iterated on prompt engineering to ensure the model delivered high-quality, medically sound, and personalized health recommendations, avoiding generic advice.
Firebase Real-time Data Synchronization: Structuring the NoSQL schema within Firebase for efficient temporal data retrieval and ensuring real-time synchronization across the Streamlit frontend proved complex. We implemented rigorous Firebase Security Rules to guarantee data privacy and user isolation, which required careful troubleshooting.
Scalable Custom Data Processing: The largest challenge was building a system that could efficiently pass structured user logs to the LLM (Gemini) and reliably process the unstructured text output into actionable UI elements.

Overcoming these issues involved deep debugging and significantly advanced our team's skills in resilient API architecture and data security.

Accomplishments That We're Proud Of

We are exceptionally proud of achieving a seamless, unified data pipeline and intelligent service layer by successfully integrating and stabilizing Streamlit, Firebase, and the Gemini API into a single, cohesive application. Our key accomplishments include:

Functional Intelligence Layer: Delivering a functional and reliable AI-driven wellness chatbot and recommendation engine using Gemini.
Real-time Secure System: Building a secure data backbone with Firebase Firestore capable of handling real-time user metrics and maintaining data integrity.
Proactive & Contextual Insight: Moving beyond simple data display to provide genuinely contextual and personalized advice that has the potential to drive meaningful behavioral change.

These accomplishments demonstrate the practical feasibility and real-world potential of VeriWell AI to improve users’ wellness journeys.

What we learned

This project provided invaluable, high-level engineering experience:

Resilient API Architecture: Gaining proficiency in handling external, asynchronous services like the Gemini API, focusing on error handling, exponential backoff, and maximizing prompt effectiveness.
NoSQL Schema Design: Deepening our understanding of efficient Firebase data modeling for complex, time-series health metrics to ensure fast and scalable data querying.
Full-Stack Cohesion: Reinforcing the importance of clear development workflows and communication for coordinating the frontend (Streamlit), backend (Firebase), and intelligence layer (Gemini).

What's next for VeriWell AI wellness coach

We envision a robust future for VeriWell AI, extending its scope to become a predictive health partner:
Predictive Wellness & Risk Assessment: Implementing advanced features that use Gemini to not only analyze current habits but also predict periods of potential high stress or low performance based on logged trends, offering proactive, early advice.
Comprehensive Nutritional Guidance: Integrating features to allow users to log meals, utilizing Gemini's multimodal capabilities (future expansion) to analyze dietary patterns and provide personalized food recommendations to address identified deficiencies or lifestyle goals.
Accessibility & Inclusion Expansion: Developing modules specifically tailored to assist individuals with various disabilities in monitoring and maintaining physical and mental health with adapted metric logging and advice.