Product Introduction
- Mai is a browser extension and pseudo API designed to extend the functionality of Meta Rayban Smart Glasses or the standalone Messenger app by integrating custom AI bots and automation workflows.
- The core value lies in enabling voice-activated interactions with third-party AI services like ChatGPT, Claude, and Perplexity through Meta Glasses’ native “Hey Meta” voice commands, bypassing platform limitations.
Main Features
- The extension allows users to send photos, messages, or video call screenshots directly to AI services via voice commands like “Hey Meta send a photo to my Foodlog” or “Hey Meta send a message to ChatGPT.”
- Real-time video monitoring captures screenshots during video calls, processes them through configured AI models, and logs responses in a dedicated viewer for analysis.
- Customizable API endpoints support integration with multiple AI providers (OpenAI, Perplexity, Claude) and text-to-speech services, with configurable API keys managed through the extension’s settings panel.
Problems Solved
- Addresses Meta Glasses’ lack of native support for third-party AI integrations by leveraging Messenger group chats as a programmable interface for voice command routing.
- Targets developers and power users seeking to automate workflows (e.g., food logging, AI-assisted research) and businesses requiring real-time video call analysis through AI models.
- Enables scenarios like hands-free image description for visually impaired users, automated customer service responses during video conferences, and instant query resolution via voice-activated AI interactions.
Unique Advantages
- Unlike official Meta integrations, Mai uses a reverse-engineered approach to map voice commands to Messenger group chats, enabling compatibility with unsupported AI services without requiring device firmware modifications.
- Innovates with a browser-based video monitoring system that captures and processes video call content in real time, a feature absent in Meta’s native SDK or competing third-party tools.
- Competes through its open-source MIT license, support for multiple AI backends (including local model deployments), and compatibility with Brave, Chrome, and Firefox browsers for cross-platform flexibility.
Frequently Asked Questions (FAQ)
- What are the requirements to use Mai? Users need Meta Rayban Smart Glasses or Messenger app access, a secondary Facebook account for bot interactions, and valid API keys for AI services like OpenAI or Claude.
- How does video monitoring work? The extension captures screenshots during Messenger video calls, sends them to configured AI models for analysis, and displays processed results (e.g., object detection logs) in the extension’s interface.
- Why is an alternate Facebook account required? Meta Glasses sync contacts and group chats from the primary account, requiring a secondary account to avoid conflicts when creating AI bot-specific group chats for command routing.
- Which browsers are supported? The extension is compatible with Chromium-based browsers (Chrome, Brave) and Firefox, with build scripts provided for each via Bun runtime and WXT framework.
- Can I use local AI models instead of cloud APIs? Yes, the API configuration supports custom endpoints, allowing integration with self-hosted models like Llama or Stable Diffusion through manual endpoint configuration.
