About Section
Dashboard
Starting new Chat with Mood.
Home page chatting with AI and making journal.

MindEase – AI-Powered Mental Health Journaling Companion

Inspiration

In a world where mental health struggles are growing but still often stigmatized, many people hesitate to express their emotions—even privately. I wanted to build something that gives users a safe, judgment-free space to process their feelings at their own pace.

MindEase was inspired by the idea of a digital companion—one that is always available to listen, offer gentle encouragement, and help track emotional well-being over time.

What It Does

MindEase is a web-based journaling companion that:

Prompts users to select a daily mood through emoji buttons
Engages in conversation based on that mood with supportive, AI-powered responses
Tracks emotional history in a dashboard with charts and mood trends
Stores chats privately, allowing users to revisit past reflections
Filters irrelevant topics and redirects users gently back to mental health themes

How I Built It

MindEase was built using the MERN stack:

MongoDB – For storing user data, chat sessions, and mood logs
Express.js – Backend API for authentication, chat management, and AI integration
React – Frontend for mood selection, chat interface, and dashboard UI
Node.js – Server logic and route handling
DeepSeek AI – Integrated via Hugging Face's inference API to generate contextual responses
React Markdown – For rendering formatted AI messages (bold, lists, etc.)
JWT Authentication – For secure access control
Chart.js – For mood tracking visuals

Key Features

Emoji-based mood selection
AI responses tailored to emotional tone (happy, sad, anxious, etc.)
Real-time chat UI that mimics human-like conversation
Typing indicator and send delay for realistic pacing
History viewer to reload past conversations
Responsive dashboard with emoji-labeled charts

Challenges Faced

Controlling AI relevance: Preventing users from using the bot for non-mental-health conversations. Solved this by prompting the AI with mood context and response boundaries.
Duplicate AI replies: Sometimes the AI responded multiple times. Implemented deduplication logic and request throttling.
Securing private messages: Used token-based auth and MongoDB schemas to ensure each chat is tied to the correct user.
UI/UX design: As developers more than designers, spent time iterating on a clean, calming interface with vanilla CSS.

What I Learned

How to build a context-aware AI interface
Managing state-driven chat UIs in React
Working with third-party AI models like DeepSeek through APIs
Importance of ethical boundaries in mental health tools
How small touches (like markdown rendering or mood-based greetings) can drastically improve UX