Patents, product, business, and technical deep-dives on AR glasses, one company at a time. Each chapter reads the same hardware story through four lenses: business (the strategic bet), product (what the device chooses to be), technical (what the patents and teardowns reveal), and application (what builders can ship today — the live demo is the proof).
An interactive 3D map of Snap Inc.'s AR-glasses patent portfolio, projected onto the 2026 Specs hardware — plus the underlying curated dataset (697 Snap documents, 65 Spectacles-specific) and the downloaded patent PDFs.
I went through Snap's Spectacles patent history. The product strategy is questionable, but the technical progression is more coherent than people realize.
Live demo → https://ar-glasses-six.vercel.app
Or open web/index.html locally — it is a single
self-contained file (data and 3D model embedded; only the Three.js library
loads from a CDN).
- Patent Map tab — a photoreal 3D model of the glasses with leader-line callouts for 6 hardware systems. Drag to orbit, scroll to zoom, double-click to reset. Click a callout for a plain-English "challenge → progress" explainer and the system's innovation cluster (one 3D node per invention, sized by patent family; click a node to load that family in the sidebar).
- Big Ideas (ELI5) tab — a click-through deck explaining which features are
a big deal and the patents behind them, now with interactive mini-demos for
the flagship inventions: play the waveguide light-path, drag the frame to watch
bend-correction cancel the wobble, slide your thumb along the finger-ruler
gesture, trigger EMG silent-speech recognition, toggle the health-sensing
face, and see how an on/off LCoS chip fakes 16.7M colors with timed bit-planes.
Each demo is grounded in the actual patent text (see
data/summaries/). - Timeline strip — every patent plotted by publication year across the 6 system lanes, showing the 2025–26 filing surge into the 2026 launch.
| Path | What it is |
|---|---|
web/ |
The interactive visualization (index.html) plus assets/spec.glb, an original 3D reconstruction of the glasses made by the author from public reference images (a copy is embedded in the HTML). This is the code that grows into the website. |
data/ |
The patent dataset: curated CSVs + data/README.md documenting scope, methodology, and caveats. |
data/pdfs/ |
40 downloaded Spectacles patent PDFs (filename = publication number). Patent documents are public records — see LICENSE scope notes. |
data/summaries/ |
Plain-English summaries of all 40 PDFs, read from the full patents (text extraction + vision for the scanned ones), grouped by area. The source notes behind the interactive explainers. |
local/ |
Not in git. Per-project working area (video productions, essay drafts, source assets, capture variants). See local/README.md. |
web/index.html is deliberately a single file today so anyone can download and
open it. Planned evolution into a proper site:
- Split the monolith: extract data JSON, styles, and modules from the inline HTML
- Host the demo (static hosting — no build step required)
- Pipeline to refresh the dataset (weekly Google Patents sweep + USPTO ODP continuity check, per
data/README.md"Keeping current") - Next chapters: the same four-lens treatment for other AR-glasses makers (Meta, Google/Android XR, Xreal, …), each with its own patent map and dataset
Contributions and issues welcome.
Code is MIT-licensed. Patent documents and bibliographic data are public
records (details in LICENSE). This is an independent research
project, not affiliated with or endorsed by Snap Inc.