Selected project / Case studyWork · Rapid prototype

Designed, built, explored

MoonView

Holographic market charts in the browser. First prototype on November 5, 2026; public launch on November 7; live subscription checkout on November 8.

MoonView share card showing a holographic candlestick chart and order book above a lunar surface
Live — moonview.appProject view
  • Rapid prototyping
  • three.js
  • WebXR
  • Stripe

Operating brief

The constraint.

Market charts are flat, and each one sits in its own window. MoonView asks what changes when the chart, the order book, and the indicators share one 3D space and one clock. The second question was speed: how quickly one builder, working with AI coding agents, can take that idea to a product people can open and subscribe to.

What the system does

  • Draws live candlestick charts, order-book depth, and indicators as 3D views in one scene.

  • Links the views, so one crosshair, one instrument, and one timeline move them together, with rewind and replay.

  • Includes drawing tools, saved layouts, and price alerts.

  • Runs in a desktop or phone browser, with a WebXR mode for VR and AR headsets.

  • Offers free and Pro plans, with accounts and subscription billing.

Architecture

How it moves.

Live market data in, tested models in the middle, linked 3D views out.

  1. Live market data feeds
  2. Tested data and indicator models
  3. Linked 3D views (three.js)
  4. Browser · phone · headset

Accounts run on Supabase with row-level security. Billing runs on Stripe subscriptions.

Build ledger

What carries it.

Interface
TypeScript · three.js · WebGL shaders · WebXR
Build & test
Vite · Vitest · more than 1,300 automated tests
Accounts & billing
Supabase (Postgres, row-level security) · Stripe subscriptions
Delivery
Vercel · content security policy · AI coding agents with human review

Current status

Delivered.

  • November 5, 2026: first working prototype.

  • November 7: public launch at moonview.app, with terms and privacy pages.

  • November 8: live subscription checkout, confirmed with an end-to-end trial sign-up and cancellation.

  • Each stage shipped with automated tests, and the accounts code went through a security review before launch.

  • Built by directing parallel AI coding agents, then reviewing and testing their work.