Inspiration
1 in 10 adults in the U.S. struggle with Compulsive Internet Use (CIU), and among teenagers across the world, that number has even been shown to be as high as 47.4% in some sample groups.
Every day, we spend hours online studying, working, scrolling, and learning. Yet we rarely stop to reflect on how that time actually affects our mental balance. This lack of clarity is where Reflectra comes in.
Generic screen-time trackers exist, but they label our digital lives as "productive" or "wasteful" without proper context. Reflectra represents something deeper, a way for people to understand why they browse what they do, how that behavior truly connects to their well-being, and how they can make positive changes.
What it does
Reflectra is a browser extension tool and AI wellness dashboard that transforms your browser history and the time you spend on it into meaningful insights.
It automatically tracks the websites you visit and classifies them by purpose (e.g. Focused Learning, Social Activities, Research, etc.) using AI semantic analysis instead of rigid prebuilt lists.
From here, Reflectra:
- Visualizes how your time is distributed across digital wellness categories.
- Generates a Digital Balance Score to show your focus-to-rest ratio.
- Lets you chat with your browsing history through an AI reflection tool powered by retrieval-augmented generation (RAG).
- You can ask things like “When do I tend to start passive scrolling?” or “Did I have a healthy mix of productivity and relaxation?”
- Access to all previous tabs through the history feature, where users can return to websites from previous sessions, which can be filtered by their categorization.
- Gives dynamic notifications on wellness tips as you're in a session
- Reflectra turns your digital footprint into a reflection of your mind — helping you browse more consciously, not less.
How we built it
- Chrome Extension (Manifest V3) – Tracks tab changes and active browsing sessions.
- Backend (ExpressJS + Supabase) – Receives tab data, timestamps, and session summaries.
- AI Categorization – Uses GPT-4o-mini for contextual classification of each website’s purpose based on titles and metadata.
- Vector Database (Pinecone) – Stores website content embeddings for RAG-based retrieval.
- Dashboard (React) – Displays time analytics, wellness visualizations, and an AI Reflection Chat powered by RAG.
- Design – Modern minimalist identity built around the Reflectra heart symbol — a union of reflection and wellness — with a calming color palette and smooth UX.
Challenges we ran into
- Contextual classification — distinguishing between “learning” and “entertainment” on the same platform (like YouTube) required nuanced AI prompts and metadata extraction.
- Privacy and efficiency — balancing meaningful insight without over-collecting user data.
- RAG implementation — ensuring relevant retrieval from vector embeddings for user reflection queries.
- Time management — designing a seamless end-to-end pipeline (extension → backend → dashboard → LLM) within hackathon constraints.
Accomplishments that we're proud of
- Built a fully functional prototype with live tab tracking, AI categorization, and a wellness dashboard.
- Designed a brand identity (Reflectra) that feels calm, modern, and AI-native.
- Implemented an AI reflection chat that summarizes user learning and habits from real browsing data.
- Created a solution that merges AI + mental health in a fresh, practical way — showing that mindfulness and technology can coexist.
What we learned
- How to integrate AI reasoning into everyday tools for human benefit, not just automation.
- The power of embedding-based retrieval to surface personalized, contextual insights.
- How design and user psychology play a huge role in making wellness technology feel helpful instead of surveillance-like.
- The importance of balance between technical feasibility and emotional resonance in health tech.
What's next for Reflectra
We plan to launch a mobile companion app that syncs across all devices.
Built With
- express.js
- google-auth
- gpt-4o-mini
- javascript
- manifest
- openai
- pinecone
- postgresql
- python
- react
- supabase
Log in or sign up for Devpost to join the conversation.