Scaling Human 实验合集：Food Scanner、Glasshole 等硬件与 AI 项目

Scaling Human Experiments

A collection of personal experiments with hardware, AI, apps, games, and local computing—built to learn, explore, and make useful things for everyday life.

Personal experimentation, not a commercial offering. This repository is maintained for learning and personal use, not as a paid product or subscription service. Individual experiments may have free App Store apps, personal sideload builds, source artifacts, or documentation only.

Humans + Machines >> Humans | Machines

Created and maintained by Srikantha Ballani (@c0smicdirt) as part of Scaling Human.

Project directory · Project briefs · Support · Contact

Explore the projects

Project	Experiment	What is available here	Project page
Food Scanner	AI-assisted nutrition from a food photo	Google Glass companion source and installation guide; iPhone source is separate	Website
Glasshole AI	Conversation summaries and fact-check overlays	Google Glass companion source and installation guide; iPhone source is separate	Website
Kids Meal Mode	Timed meal prompts on Fire TV / Android TV	Project documentation and personal/family sideload download; source is private	Website
Book Cricket	A nostalgic page-flip cricket game	Links to the separately maintained project; source is not included here	Website
Local AI game experiment	An offline model builds a browser platformer	Generated game source, README, and test harness	Website
Smart Ring Local	Direct Oura Ring 4 Bluetooth connection to iPhone	Project scope and links; native app and hardware validation pending	Website
Smart Strap Local	Direct WHOOP 5.0 Bluetooth connection to iPhone	Project scope and links; native app and hardware validation pending	Website

How the collection is organized

Each experiment has a project brief that links to its existing code, documentation, downloads, and public page. Not every project has public source in this repository. GitHub Projects boards can track each app's issues and milestones while the code and release assets remain in their existing locations.

This is a collection rather than one installable application. Start with an individual project's page for its scope, status, installation options, and limitations. The existing source-available license governs the published repository; it is not an unrestricted open-source license.

The website maintains the public project pages, support and privacy URLs. This repository keeps shared artifacts, guides and releases. Private app implementations remain separate. The organization overview connects the repositories; the maintainer's profile covers other interests.

Published app, document and download paths stay stable when navigation changes. See the website's link policy before moving files, changing release assets or renaming a repository.

Google Glass companion details

Food Scanner and Glasshole include optional wearable companions. Their source, setup instructions, supported providers, and screenshots are below.

Food Scanner & Glasshole — companion setup and screenshots

Companion Experiments

The Food Scanner and Glasshole projects include optional Android companions for Google Glass Explorer Edition. This section documents those two companions; other experiments have their own project briefs and installation paths.

Food Scanner captures a food image and sends it to the companion iPhone app for nutrition analysis, then displays the result on the companion display.
Glasshole AI is an AI-powered conversation fact-checker. It starts and stops real-time conversation capture and displays summaries and fact-check overlays returned by the companion iPhone app after analysis.

Use Glasshole only with the consent of everyone being recorded and in compliance with local laws and regulations where it is used.

The free iPhone apps are distributed separately through the Apple App Store. The iPhone apps also support standalone operation without companion hardware.

Repository Contents

companions/
  foodscanner-glass-companion/
  glasshole-glass-companion/
docs/glass/
assets/images/
scripts/

The private iOS source, App Store submission materials, research papers, and internal development notes are not published in this repository.

Supported AI Providers

The iPhone apps let the user choose the processing provider:

Glasshole: Apple Intelligence, xAI Grok, OpenAI, Anthropic Claude, Google Gemini, Ollama, PocketPal, or a custom OpenAI-compatible endpoint.
Food Scanner: Local iPhone Vision, xAI Grok, OpenAI, Anthropic Claude, Google Gemini/Gemma, Ollama vision models, PocketPal, or a custom endpoint.

Apple Intelligence in Glasshole requires an iPhone 15 Pro model or an iPhone 16 model or later, with Apple Intelligence enabled. Other supported iOS 26 iPhones must use a configured cloud, local-network, PocketPal, or custom provider. Food Scanner's default Local iPhone Vision mode works without Apple Intelligence or an API key.

Get API keys or local runtimes:

Install

See the Google Glass XE Installation Guide for ADB installation, networking, and how to launch companions with a touchpad tap on the fact check or scan food Glass commands.

Do not rely on saying “Ok Glass” — use the touchpad.

Product pages (download APK + link to this guide):

The companion apps target Android 4.4.4 / API 19 and are sideloaded.

Screenshots

Food Scanner

Glasshole

Status

Experimental. AI nutrition estimates and fact-check results can be wrong and must not be treated as medical advice or authoritative fact.

Source is available under the Scaling Human Source-Available License. Redistribution, commercial use, hosted derivatives, and app-store publication require written permission.