Inspiration
Mental wellness tools often fall into two extremes: passive static digital journals that offer zero feedback, or overwhelming multi-step therapy platforms. During high-stress moments—like back-to-back academic exams or late-night burnout—users need immediate clarity, non-judgmental validation, and rapid actionable resets. We built MindLoom AI to bridge this gap by transforming raw, stressful thoughts into structured emotional reads and targeted behavioral micro-habits within 60 seconds.
What MindLoom AI Does
MindLoom AI is a privacy-focused mental well-being companion designed to untangle daily cognitive overload:
- Voice-to-Text & Rapid Reflection: Users can express their thoughts instantly using context tags (
#AcademicStress,#Overwhelmed) or hands-free voice input via the browser's Web Speech API. - Gemini-Powered Sentiment Engine: The app processes input in real time to calculate primary emotion scores, sentiment confidence metrics, and compassionate behavioral insights.
- Interactive Micro-Resets: Instead of static advice, MindLoom provides an interactive guided breathing and grounding timer directly on screen to regulate stress immediately.
- 14-Day Analytics Dashboard: Tracks mood balance trends, reflection streaks, and emotional distributions across 14 days to provide users and therapists with actionable longitudinal data.
- Responsible & Safe AI: Integrates visible medical disclaimers to clearly separate wellness self-care from clinical therapy.
How We Solved Core UX & Technical Challenges
1. Eliminating Input Friction with Speech-to-Text Integration
- Problem: Users experiencing acute burnout or emotional exhaustion often find typing out long journal entries mentally draining.
- Solution: We implemented the native browser
WebSpeech API(SpeechRecognition). When users tap the microphone icon, spoken audio is captured statelessly and transcribed in real-time directly into the reflection buffer, allowing fluid, hands-free journaling.
2. Preventing Static AI Responses & Delivering Targeted Interventions
- Problem: Typical mental health wrappers output generic box-breathing prompts regardless of whether a user suffers from study paralysis or late-night anxiety.
- Solution: We structured our Gemini API prompt architecture to parse both emotional tone and context tags (
#AcademicStress,#Burnout,#Anxious). The UI dynamically renders custom micro-tasks—ranging from 5-minute Pomodoro action dumps for study stress to parasympathetic down-regulation for anxiety.
3. Turning One-Time Utility into Sustainable Longitudinal Retention
- Problem: 80% of hackathon AI wrappers fail because they operate statelessly without long-term value.
- Solution: We built a 14-day tracking dashboard using
Recharts. Every reflection updates emotional distribution metrics, streak counters, and weekly stability trends so users can observe their mental baseline over time.
Tech Stack
- Frontend UI: React, Tailwind CSS, Lucide Icons, Recharts
- AI Processing Engine: Google Gemini API
- Audio & Input: Web Speech API (
window.SpeechRecognition) - Development & Prototyping: Lovable platform
What's Next for MindLoom AI
- Therapist Data Export: Generating downloadable PDF emotional baselines to share directly during clinical sessions.
- Offline Storage Persistence: Full local-first encryption using
localStorage/IndexedDB. - Customized Audio Soundscapes: Binaural beats matched to detected emotional states.
Built With
- google-gemini-api
- lucide-react
- react
- recharts
- tailwind-css
- typescript
- web-speech-api
