Mind Laundry

Inspiration

This project was inspired by those nights when worries and anxiety keep you awake long after the day is over. When unfulfilled worries grow, it sometimes feels as though the worries are controlling you, rather than the other way around.

Mind Laundry offers an experience where you write down these worries, wash them, and hang them out to dry. I wanted to transform worries from an overwhelming force into tangible objects that you can hold in your hands and set aside for a while.

What it does

Mind Laundry is an emotional relaxation app that treats worries and anxieties as pieces of "laundry." When you type out lingering thoughts, the sentences transform into laundry that you can dip in water, scrub on a washboard, rinse, and hang out to dry on a clothesline.

Pressing the send button summons a breeze that scatters the laundry into tiny particles of light, forming a constellation in the night sky. This doesn’t mean the worries have vanished completely; rather, it symbolizes that you have dealt with them as much as you can for today and can now let them rest.

The app does not analyze emotions, provide correct answers, or utilize mechanics like questions, scores, or rewards. Instead, it simply helps users unpack and process their thoughts at their own pace before ending their day.

How we built it

Mind Laundry was built using Unity’s UI system, TextMeshPro, and the Input System to seamlessly connect mobile touch interactions with text animations.

The core technical challenge was turning typed sentences into tangible objects, rather than just readable text. I manipulated TextMeshPro’s mesh vertices at runtime so the sentences bend, sag, and sway like fabric. The physics are configured to convey inertia and restoring forces based on drag direction and speed. Actions like soaking, scrubbing, rinsing, hanging, and wind effects were all directly tied to user inputs.

Visually, I used a black night sky, white lines, and a pencil-drawn interface to evoke the feel of a moving sketch. While I utilized LLMs during the development process, I deliberately excluded LLM integration, sentiment analysis, and auto-summarization from the final product to ensure the app never attempts to interpret or "solve" the user's worries for them.

Challenges we ran into

My biggest challenge was implementing natural physical behaviors—like the sagging, folding, and swaying of a towel—with the help of an LLM. Initially, I tried directly modifying the vertices, but the "towel" felt stiff, and the movements were erratic, failing to simulate real fabric.

Thanks to advice from a client developer (my younger sibling), I reframed the problem from simple vertex manipulation to a rigging approach. By creating a grid of bones within the towel mesh, simulating physics to move these bones, and having the mesh and text follow the resulting surface, I was able to achieve highly natural movements.

Accomplishments that we're proud of

I am most proud of making the user's written sentences the literal center of the interaction. Instead of converting worries into avatars or abstract items, the actual text containing the worry gets wet, sways, and hangs on the clothesline.

Through a cohesive metaphor—typing, washing, hanging, blowing in the wind, and turning into constellations—I successfully conveyed the sensation of physically handling and letting go of one's anxieties. I created a self-contained experience where the mere act of taking out and processing one's worries is enough, completely free of analysis, advice, or scoring mechanics.

What we learned

This project taught me that while LLMs are incredibly useful, domain knowledge is essential to achieve effective results. The LLM excelled at writing code rapidly and prototyping various approaches. However, without a foundational understanding of rigging and cloth physics, my prompts often led to repetitive iterations of flawed methods rather than fundamental solutions.

Once I accurately defined the problem and proposed the correct architectural structure based on domain knowledge, the LLM drastically accelerated implementation and repetitive tasks. The key takeaway is that you cannot outsource judgment to an LLM; it is most powerful when humans set the direction, validate the output using their expertise, and collaborate with the tool.

What's next for Mind Laundry

For my next step, I plan to expand this one-time relaxation experience into a quiet diary for reflecting on past worries. Previous "laundry" will be archived as dated clotheslines or stars, serving as a gentle reminder that the user has successfully endured and moved past previous anxieties.

To protect privacy, users will be able to choose whether to save the original text or only retain unidentifiable traces, such as dates or the position and brightness of a star. I also plan to introduce various washing methods—such as hand-washing, a drum washing machine, or using a traditional laundry bat—each with its own unique animations and sound effects.

Ultimately, my goal is not to build an app that magically solves worries for users, but to create a safe space where users can repeatedly reassure themselves that they have the strength to manage their own anxieties.

"I have overcome my past worries. I can handle today's as well."