BookCrafter

• 

  Poster

Aura: Your Wellbeing Copilot

Aura is not just another life-tracking app; it's a proactive, AI-powered copilot for your holistic wellbeing. It operates on a simple yet profound principle: the different domains of your life are deeply interconnected. By understanding how your sleep affects your productivity, how your social interactions influence your mood, and how your daily habits shape your future, Aura transforms raw data into a clear path for intentional living.

The Inspiration: From Data Chaos to Predictive Clarity

I was inspired by a sense of digital fragmentation. Many of us use a dozen different apps to manage our lives: a calendar for our schedule, a fitness tracker for our health, a to-do list for our tasks, a budgeting app for our finances, and a notes app for our thoughts. Each app is a silo, holding a single piece of the puzzle.

The core question that sparked Aura was: What if these silos could talk to each other?

• What if an app could see that three nights of poor sleep and a packed meeting schedule would lead to a productivity crash on Friday?
• What if it could correlate a decline in social interactions with a dip in your mood and proactively suggest reconnecting with a friend?
• What if, instead of just reacting to life, we could have an intelligent copilot that helps us anticipate challenges and seize opportunities before they arrive?

I wanted to build a system that moves beyond simple historical tracking to offer true predictive intelligence, empowering users to become the architects of their own wellbeing, not just recorders of it.

How It's Built: A Modern, Secure, and Intelligent Stack

Aura was fully designed and built in bolt.new.

• Frontend:

  • The user interface is crafted with React 18 and TypeScript, ensuring a robust and type-safe codebase.
  • We used Vite for a blazing-fast development experience.
  • Styling is handled by Tailwind CSS, allowing for rapid, utility-first design that adheres to a premium aesthetic.
  • Smooth, purposeful animations are powered by Framer Motion, and beautiful, interactive charts are rendered with Recharts.

• Backend & Database:

  • The entire backend is powered by Supabase, which provided an incredible developer experience.
  • PostgreSQL serves as our powerful relational database, with a comprehensive schema of over 10 interconnected tables to model a user's holistic life.
  • Supabase Auth handles secure user management, while Row Level Security (RLS) is enabled on every table from day one, guaranteeing that users can only ever access their own data.
  • Supabase Edge Functions (written in Deno/TypeScript) are the workhorses of our AI engine. They aggregate user data on the fly and run our predictive models.

• AI & Integrations:

  • The Foresight engine uses a combination of trend analysis and correlation models within an Edge Function to generate predictions for stress, productivity, and mood.
  • The AI Video Coach feature is designed for integration with the Tavus API.
  • Initial data is simulated to represent real-world feeds from Apple Health, Google Fit, Oura Ring, Strava, Google Calendar, Todoist, and Plaid.

What I Learned: Beyond Code and into User Empowerment

Building Aura has been a journey of immense learning, both technically and philosophically.

1. Privacy is Not a Feature; It's the Foundation: Handling such sensitive personal data hammered home the importance of a privacy-first architecture. Implementing and testing Row Level Security (RLS) in Supabase from the very beginning was the most critical lesson.
2. The Power of a Unified Backend: Using Supabase streamlined development immensely. Managing auth, database, and serverless functions in one place allowed me to focus on the core product logic instead of juggling disparate services.
3. Explainable AI is Crucial for Trust: It's not enough for an AI to make a prediction. To be a true "copilot," it must explain why. This led to the creation of the "Key Triggers" feature in Foresight. Showing a user that their stress is predicted to rise because their sleep is down and their meeting load is high builds trust and empowers them to take targeted action.
4. The Art of Data Visualization: Turning complex correlations and time-series data into simple, beautiful, and understandable charts (like the Wellness Radar and Risk Timeline) is a design challenge as much as a technical one. The goal is to provide insight, not overwhelm.

Challenges Faced & Solutions

1. Designing a Holistic Data Model: The biggest initial challenge was designing a database schema that could capture the richness of a person's life without becoming impossibly complex. The solution was to categorize data into distinct but interconnected tables (health, tasks, social, etc.) and use user_id and dates as the primary linking mechanisms.

2. AI without a Massive Infrastructure: Building a sophisticated predictive engine without a dedicated team of data scientists and a massive ML pipeline was daunting. I overcame this for the MVP by starting with simpler, yet effective, models inside Supabase Edge Functions. These functions run custom logic, trend analysis, and rules-based correlations that provide surprisingly accurate initial insights, proving the concept's viability before scaling to more complex models.

3. Performance with Large Datasets: Fetching and aggregating data from 10+ tables for the dashboard and Foresight could be slow. The solution involved careful query design, creating database indexes on user_id and date columns, and architecting the data aggregation logic in the Edge Function to be as efficient as possible.

4. Avoiding a "Black Box" UX: The initial designs for Foresight felt too much like a magic eight ball. The challenge was making the AI transparent. The solution was to break down insights into three clear parts: the Prediction (what will happen), the Key Triggers (why it will happen), and the Interventions (what you can do about it). This structure transformed the feature from a novelty into a truly usable tool.

Built With

• gemini
• nextjs
• typescript