Mellow

Light Mode: Homepage with journaling, Growth Areas, Shower Thoughts, and Voice memo; Main journaling page
Insights; Calendar view
Dark Mode: Homepage; Calendar page with sticky note and image attached

Mellow: Project Story

Inspiration

Non-journalers still reflect — just unintentionally.
They vent about stress, think about goals, notice patterns, and want to improve — but those thoughts disappear because there’s no lightweight system to capture and make sense of them.

Mellow was inspired by that gap.

What if journaling wasn’t a hobby, but a tool for lifestyle-maxxing?
What if reflection felt effortless — and actually helped you improve?

What it does

Mellow is journaling for lifestyle-maxxing non-journalers — an app that understands you and motivates you to grow.

Instead of asking users to journal for journaling’s sake, Mellow:

Gives simple daily prompts so users never face a blank page
Supports quick, low-commitment entries
Stores short-term reflection data (last 1–2 weeks)
Analyzes what users write to identify growth areas
Surfaces clear, actionable improvements users can work on

For example, if a user consistently writes about feeling overwhelmed by work, Mellow doesn’t just summarize that — it highlights “Managing Work Stress” as a growth area and suggests a realistic adjustment.

Mellow turns reflection into direction.

How we built it

We built Mellow with speed, clarity, and iteration in mind.

Frontend

Designed the UI using v0 to rapidly prototype flows and visual hierarchy
Built the mobile experience with React Native (Expo)
Focused on a calm, modern interface optimized for non-journalers:
- Home dashboard with daily prompt
- Quick entry flow
- Growth Areas section that highlights what to improve next

Backend

Used Supabase for authentication and data storage
Modeled journal entries and AI-generated insights in Postgres
API calls with Dedalus Labs
Implemented Supabase Edge Functions to:
1. Fetch a user’s recent entries (short-term memory)
2. Detect recurring patterns using AI
3. Store structured “growth areas” back into the database

Challenges we ran into

Avoiding “self-help cringe”: The app needed to feel practical and motivating, not emotional or preachy.
Scoping AI responsibly: We intentionally limited memory to short-term data to avoid overreach while still delivering meaningful personalization.
Turning reflection into action: It was harder than expected to phrase insights in a way that felt helpful rather than judgmental.
Time pressure: Balancing backend logic, UI clarity, and AI behavior within a hackathon timeframe required strict prioritization.