MindDump

Dashboard
Weekly Insights
Focus Mode
Home

Inspiration

The idea for MindDump came from a common, frustrating feeling that is mental clutter. I noticed this pattern especially among students, professionals with ADHD, and even myself. As someone who juggles thoughts, deadlines, dreams, and doubts all at once, I often found myself forgetting tasks, misplacing ideas, or getting overwhelmed by thoughts that had no place to go. Task managing apps felt too rigid and robotic. Notes apps were a mess. So, I imagined a place that simply said: “Dump it here. We’ll take care of it.” That’s where MindDump started, a mental inbox for everything on your mind, where thoughts could be smartly organized without judgment. Moreover, I was always inspired to create a productivity app since childhood, thanks to Bolt.

What it does

MindDump is a web app that gives your cluttered thoughts a home — and then helps you make sense of them.

At its core, it allows users to freely write whatever is on their mind, whether it's a random idea, a personal feeling, or a to-do item. Once a thought is "dumped," the app intelligently classifies it using the Gemini API — sorting it into meaningful categories like Tasks, Personal Reflections, Shopping Lists, Reminders, or Random Thoughts.

But MindDump doesn’t stop there.

If the thought is actionable, it’s treated like a task, and users can:

Set it to Focus Mode, where a Pomodoro timer helps them complete the task in a distraction-free space with sound alerts and a visual progress tracker.
Mark it as “done” when completed or like it as a form of self-validation.
Update the status, move it between categories, or edit/delete as needed.

How we built it

This project was built entirely during the Bolt.new Hackathon using prompt-based development, with intense focus on React, Supabase, and Gemini API. Here’s a snapshot of the journey:

Started with a React-Tailwind template
Implemented a basic thought box (dumping feature) with auto-save to Supabase
Authentication via Supabase (with conditional UI for dumping)
Prompt-engineered Gemini to smartly categorize thoughts (shopping, personal, etc.)
Fixed issues where dumps without login still worked
Built a dashboard showing categorized thoughts with filters (pending/done/urgent)
Added AI urgency detection for urgent tasks with red animation alerts in dashboard
Designed and integrated a Pomodoro-based Focus Mode with sound alerts, allowing users to move between tasks using visible < and > navigation buttons
Developed a Weekly Insights feature showing data visualizations (bar charts, graphs) of user activity with motivational messaging
Enabled functions like editing, deleting, marking done/undone, moving tasks between categories
Implemented thought vs task classification, for example, "life is boring" is recognized as a thought (non-actionable) and treated differently than a task
Added a complete dark mode toggle that reflects across all UI components
Designed an animated shuffle-to-organized word transition for the homepage
Created a realistic footer by attaching the Bolt badge and repositioning text as required

Challenges we ran into

Throughout this project, we encountered both expected and surprising challenges, especially given our AI-heavy workflow:

Supabase email authentication links failed due to localhost redirection issues during early testing.
We initially tried using DeepSeek API for categorization, but it lacked reliability in identifying nuanced thoughts leading to a switch to Gemini.
Categorization sometimes resulted in inappropriate statuses, like marking non-actionable thoughts (e.g., “life is boring”) as “pending tasks.”
The first design for red alerts on urgent tasks was visually aggressive and clashed with the dashboard UI, so we redesigned it with smoother animations.
Dark mode was only partially applied initially (only headers), and extending it across the entire app needed manual overrides.
The “blur” effect on the “Dump it” button before login was unreadable; we had to fine-tune it to maintain both aesthetics and clarity.
Gemini also struggled with edge cases and sometimes misclassifying thoughts was handled by us.
Achieving clean, responsive spacing across components while maintaining visual hierarchy took multiple layout passes.

Accomplishments that we're proud of

Smart task categorization using Gemini API
Clean distinction between "thoughts" and actionable "tasks"
Fully responsive and dark-mode compatible UI
Pomodoro timer with audio alert and focus navigation
Creative homepage animation showing word “ORGANIZED” being shuffled then aligned
Integrated Bolt Badge with styled footer
Proper authentication gating with memory option (“remember me”)
Advanced dashboard features (edit, delete, status toggle, category switch)
Weekly analytics & mood-based motivational summary

What we learned

Prompt engineering needs iteration, one wrong word can misguide AI
Always clarify UI intent, Bolt sometimes misinterprets instructions
Simpler UI wins when handling complex data
Categorizing human thoughts is not rule-based; it requires true AI understanding
Test every state of app like unauthenticated, typo-heavy, and edge cases