Loading.

Seen through the headset at night: a gold web of spheres and lines radiating from one bright node across a dark attic room, with a blue web around it and the search window floating to the left

Status

Ongoing

Team

Solo project

Platform

Apple Vision Pro

Built with

Claude, Gemini, ChatGPT

This project started after one of my many passion-fueled YouTube deep dives, when I came across a graph of Wikipedia, a video that blew up years ago. That idea fascinated me. To me, Wikipedia stood in for all of human knowledge, and here was a new way to see it, and most importantly, how every piece connected to one another.

Even back then, from high school on, I wanted to walk through that graph and feel the wonder and connection in it. It was almost like looking up at a starry night and feeling that sense of awe, but around human knowledge. Building something I could actually walk through, something that made the idea concrete, was not something I could even envision.

Then I was lucky enough to explore it further in Virtual People at Stanford. I got to really flesh out what it could be, writing up ethics reports and the overall idea of the project, and editing a video by hand to bring the idea into the world Watch ↓. But it was just a dream, and I definitely couldn't build it myself.

That changed after I jumped into RBV, Reality Becomes Virtual, a project at the Virtual Human Interaction Lab. Seeing how powerful AI agents had become, I wanted to find out if the idea I had built up could become real. After many more explorations into APIs, the vector spaces AI uses to hold ideas, the sources of data I could pull from, and the 3D visualizations that already exist, it came to life Watch ↓.

Our current search strips facts of their surroundings, and I wanted something that puts them back. I wanted the feeling of going to the library as a kid for one book and finding, right next to it, a book that's even better.

That is what I have been creating here: a web of all human knowledge that surrounds you like a starry night, free for you to explore.

Loading's Journey

How it was born

2025

I.The start

The concept video, from way back in fall 2025.

DESIGN DOCUMENTThis design template is intended to support your design process, not restrict it. We encourage you to engage with as many of the following design elements as possible within the assignment timeframe, but we recognize that covering all of them may not be possible or relevant for all students.

ELEVATOR PITCH | 3–5 sentencesDescribe the VR experience’s core concept

Doing research on any topic is a wonderful journey of exploration… but without the epic experience of being an explorer. Imagine your exploration of data, articles, videos, etc. was as impactful an experience as those childhood hunts for dragon treasure. That’s what Web Research VR gives you.

RATIONALE | 3–5 sentencesExplain why it is necessary to use VR for this experience

We’ve had two dimensions to explore data and research for some time, however several problems arise with the current system.
First, there’s a lack of spontaneous discovery, which I’d argue arises from the loss of the spontaneous interaction during research in, say, a library. In a library you can go looking for a certain book, but in asking the librarian they may suggest another similar book, or in walking to the shelves to pick up the book you want, pass a display that touches on the very thing you were researching. It’s even possible that right next to the book you wanted there’s another whose contents cover the very thing you are looking into. However, not only does it take effort to go to the library, but not every library has content which relates to your research, and it’s impossible to create a library that reflects your interests for every theme you want to dive into.
Second, even if you could create a physical library, there are some affordances which would make a library more wonderful for research. Imagine if every book with references to other books had a small line connecting one another, so you could see at a glance if maybe one source was highly referenced and you should take a look at it.

ETHICAL | 3–5 sentencesHighlight any ethical considerations or potential concerns associated with implementing your VR experience

Privacy: Privacy will be important beyond the standard VR worry of all of the cameras recording your space and those around you. This is because the system will be actively pulling in and making inferences about objects in your space in a very humanly understandable format, making the data more immediately personable than other data points (such as history) which individuals seem to not really care about regardless of their actual importance.
Whose data?: This is connected to privacy in a sense, but gets more at the idea of property. If my VR experience allows you to “store” an object you see in reality quickly, what are the implications? Are there some objects/things which should not be stored in a virtual world in certain ways? I don’t know the answers yet or where this would come up, but it feels like a potentially worrisome prospect.
Systemic connection bias: If the research system which finds connected sources/topics to what you are looking into is trained/programmed based on what is out in the world, there is a potential for super loud (but incorrect) points to be highlighted, worsening echo chamber effects, as people would feel like they are doing amazing research while actually just sticking inside their bubble.

Rough drafts of my thinking for the project, also from fall 2025.

September to October 2026

II.First test

Seen through the headset in daylight: a web around a Sergey Brin node with his card open, and beside it a To check panel listing tests, each with Pass and Fail buttons — A newer build: the new immersion button, and photos in the orbs.

Two webs, one blue and one green, overlapping in a dark room lit by string lights — First attempt at more than one web at once.

A web centred on a ringed blue sphere for the United States, with its card open beside the search window, in a lit room with a pink couch — The USA web.

I design, test and merge. AI agents write the code.

October 2026

III.How I build it

Loading gets made in loops: dream it up, build it, test it, fix it, and go again.

1 · Ideate

It starts in a variety of ways: drawing on paper, long voice messages, talking with people, thinking out loud. Inspiration comes from classes and from what's going on in the world. This project, for instance, was reignited by my AI governance class at Stanford. Learning how AI uses embeddings showed me there's a whole science of placing concepts in a multidimensional space, yet we still explore knowledge on flat, two-dimensional screens. That brought me back to the project from a year ago, in the video above. The knowledge is already spatial; there just had to be a way to experience it.

I share everything with Claude, Gemini and ChatGPT. I describe roughly what I picture and where I imagine the data coming from, and ask for three different interpretations as real HTML mockups: 2D animations, layouts, or sketches of how it would pull in data. Then I refine and narrow it down.
2 · Build

Building a VR app with AI has its own problems. AI code means lots of hiccups and errors along the way. But with the right process, you can iterate really quickly and polish them down.
3 · Test on the headset

As one person, it's hard to stop and write down what's wrong with a headset on while you're walking through the experience. So a note-taker, Granola, writes down everything I say as I go. It's been a fabulous way to capture the process.
4 · Go through the notes

Last, I go through the notes and add thoughts or extra points I've uncovered. Then the loop starts again.