P0 Foundation and local setup 5/5 done
P0.1 Run the app locally
P0.2 Provider keys
P1 Textile data layers 8/9 done
P1.1 Curated datasets
P1.2 Wiring into the app
P1.3 Seen working in a browser
P2 Live data feeds 4/7 done
P2.1 Live ships from AISStream
P2.2 Active fires from NASA FIRMS
P2.3 Weather layers stay available
P2.4 Show cargo and container ships only
P2.5 Live weather and crop signals over cotton regions
P2.6 Port congestion estimate from ship dwell time
P2.7 Live data reachable from the public site while the Mac is on
P3 Voice control 3/6 done
P3.1 Upstream improvements merged
P3.2 Voice runs on a ChatGPT sign-in, no API credits
P3.3 Textile layers addressable by voice
P3.4 Voice can answer textile questions
P3.5 Textile-specific voice instructions and first-run copy
P3.6 Free voice fallback that needs no OpenAI account
P4 Quality and release 3/5 done
P4.1 Full unit test suite passes
P4.2 Code published to GitHub
P4.3 Upstream MIT licence and credit preserved
P4.4 Dataset accuracy review
P4.5 Textile branding in the app shell
P5 Documentation 4/6 done
P5.1 Phasedoc cockpit running
P5.2 Betadoc site written
P5.3 Betadoc published on a static host
P5.4 Narrated demo video
P5.5 Studio-quality voiceover for the demo video
P5.6 Demo video can be rebuilt from a clean checkout
P6 Future roadmap 0/7 done
P6.1 Cloud twin so the demo is up 24x7
P6.2 Real factory and mill locations from OpenStreetMap
P6.3 Cotton price and trade-flow feeds
P6.4 Supply-chain disruption alerts
P6.5 Saved views and shareable briefings
P6.6 Several public demos share one tunnel port through Caddy
P6.7 Push-to-deploy with GitHub Actions
In plain language
Getting the app running on a laptop. This phase is about the basics: copying the code, installing what it needs, and adding the free access keys that unlock maps, ships, fires, traffic and street photos. Nothing here is textile-specific yet, but everything later depends on it.
This is the heart of the textile product. We added four new map layers that show where cotton is grown, where clothes and fabric are made, which ports they ship through, and the main sea routes between them. The data is a hand-curated starter set, so it does not update by itself. The remaining work is proving it looks right in a real browser.
Some information can be live. Ship positions, wildfires near farms and weather already come in from public services, and the public website relays them from the builder's own computer, so they are only live while that computer is on. The open work is making that live information textile-aware: only cargo ships, weather over cotton fields, and a rough read on how crowded a port is.
Talking to the globe. You can say 'show cotton regions' or 'take me to Tiruppur' and it responds. It currently works through a ChatGPT sign-in because the paid developer account has no credit. That route is experimental, so the open work is a free fallback and teaching voice about textile questions.
Making sure what we built is trustworthy and shared properly. The automated checks all pass and the code is on GitHub. Still open: double-checking every place name and coordinate against a second source, and re-skinning the app so it reads as a textile product while crediting the original authors.
Telling the story. This phase covers the internal progress window you are looking at now (the phasedoc) and the public website and written guide for outsiders (the betadoc). The website, the app and this page are on the internet at their own agent01.online addresses; a 64-second demo video is on the betadoc with a computer-generated voice; a studio-quality voice is still to do.
Where the product could go next. Hosting a copy that is always online, loading real factory locations at scale, adding cotton price and export data, raising alerts when a storm or fire threatens a hub, and letting teams save and share briefings. None of this is started, so every item here is honestly open.
Plan of record
Textile Globe — plan of record
The Textile Globe is a fork of an open-source 3D world monitor called God's Eye View (MIT licence, by Bilawal Sidhu). We changed it to show the textile industry: where cotton grows, where fabric and clothing are made, and how goods ship between them.
P0 Foundation and local setup
The app runs on a laptop with a single command. We copied the code, installed its dependencies, and added five free access keys (called API keys, which are passwords that let our app ask outside services for data): Cesium ion for 3D maps, AISStream for ship positions, NASA FIRMS (Fire Information for Resource Management System) for wildfires, TomTom for traffic, and Mapillary for street photos. The keys live in a private file that is never uploaded to GitHub.
P1 Textile data layers
Four new map layers were added: Cotton Regions (28 growing areas), Textile Hubs (41 manufacturing clusters), Textile Ports (36 container ports) and Trade Lanes (11 sea routes). They come from curated files stored with the code, written from general knowledge, so they are approximate and do not update on their own. The layer list was trimmed so unrelated layers (flights, military, satellites and others) are hidden but not deleted. The automated tests and the production build pass; what is still missing is a recorded walk through a real browser showing each layer draws and each marker opens its information card.
P2 Live data feeds
Ships (from AISStream, the Automatic Identification System that vessels broadcast), wildfires and weather are live, and they reach the public website through a small read-only server on the builder's own computer. That means the live layers only work while that computer is on. Three textile-specific upgrades are open: show only cargo and container ships, show weather over cotton regions, and estimate port congestion from how long ships sit near a port.
P3 Voice control
A microphone button lets you speak commands. The paid OpenAI developer account had no credit, so voice runs on an experimental ChatGPT sign-in route that the upstream project added. It works, but OpenAI has not promised that route will keep working. Textile layers are registered with the voice system, so you can say "show cotton regions". Open work: answering questions like "which hubs are in Bangladesh", textile wording in the first-run screen, and a free fallback that needs no OpenAI account (an on-device voice option exists upstream).
P4 Quality and release
All 6,214 automated tests pass and the code is on GitHub at nishadiyer/World-Monitor. The original MIT licence and credit to the upstream author are kept. Still open: an independent accuracy review of every place and coordinate in the datasets, and replacing the remaining generic branding with textile branding while still crediting the original project.
P5 Documentation
Two documents tell the story. This phasedoc is the internal, honest progress window for the builders. The betadoc is the public website and written guide for outsiders. Both are written and published on Vercel at textile.agent01.online (betadoc) and textile-phasedoc.agent01.online (this page); the live app is at textileapp.agent01.online. A 64-second narrated demo video is embedded in the betadoc; its voice is computer-generated, and a studio-quality voice is still open.
P6 Future roadmap
Nothing here is started. Ideas, in rough order: host an always-on copy behind a PIN; load real factory locations from OpenStreetMap at scale; add cotton price and export figures; raise alerts when a storm, fire or port closure threatens a hub; and let teams save and share briefings.