JellLight

Most burnout isn’t loud. It’s subtle, gradual, and easy to miss. JellLight listens to the quiet signals and reflects them back, so you can understand yourself before things feel too heavy.

Comment

🪼 Inspiration

Burnout doesn’t usually show up all at once. It builds quietly through late nights, long days, and the habit of telling ourselves we’ll rest later. Most of the time, we don’t notice how much we’re carrying until it already feels overwhelming.

As students and builders, we wanted a gentle way to pause, check in, and understand how our habits and emotions change over time. JellLight 🪼 was created to be that small moment of awareness — a calm space that helps people notice early signs of burnout and feel supported before things become too heavy.

💡 What it does

  • 📝 Daily check-ins with four fixed questions and rotating questions throughout the week, followed by a short, supportive response
  • 🧠 Mental health fun facts shown at each check-in to encourage learning in a light, low-pressure way
  • 📅 Weekly reflections with a burnout score, a plain-language summary, gentle actionable steps, and an AI-generated sticker representing the week’s mood
  • 📊 “What I’ve Noticed” page that visualizes trends in sleep, stress, and mood over time
  • 🌸 Bloom directory that helps users find therapists and mental health resources when they feel they need extra support

🛠️ How we built it

JellLight was built with a modern, full-stack architecture designed for clarity, scalability, and responsible AI use.

🎨 Frontend

Built with React + TypeScript, focusing on a calm, accessible interface with soft visuals, clear hierarchy, and smooth interactions.

🧩 Backend

Built in TypeScript, responsible for data storage, trend analysis, burnout logic, and APIs that connect the app end-to-end.

🤖 LLM Integration

Large Language Models are used to translate patterns into human-friendly insights, weekly summaries, reflection prompts, and AI-generated mood stickers. Instead of predicting or diagnosing, the LLM focuses on explanation, tone, and emotional clarity, with strict guardrails to ensure all outputs remain supportive and non-clinical.

⚠️ Challenges we ran into

  • Keeping the tone right while using AI responsibly, especially in a mental health context
  • Connecting APIs smoothly between frontend and backend
  • Defining a strong MVP that balanced meaningful UX with what was realistically achievable in 24 hours

📚 What we learned

  • Tone and wording matter as much as features in mental health tools
  • LLMs are most effective when used to support reflection, not replace judgment
  • A well-scoped MVP leads to a stronger, more meaningful product

🚀 What’s next for JellLight

  • Personalize insights using longer-term patterns
  • Expand the Bloom directory with localized resources
  • Enhance weekly reflections with richer AI-generated visuals
  • Explore privacy-first and on-device AI approaches

JellLight aims to be a small, steady light , helping people notice burnout early and feel less alone while doing so.

Built With

Share this project:

Updates