Meta open sources the code to build your own Muse AI companion gadgets
Meta just open-sourced the core code that powers its Muse AI agent, and now you can build your own physical gadgets around it.
What you get
Meta released the software stack that runs Muse, its conversational AI agent. The code is available so anyone can load it onto custom hardware. Meta suggests several reference builds: putting Muse on a color E Ink display for showing reminders, or embedding it into an HDMI stick to display on a large screen.
This is different from just calling an API. You get the actual agent runtime, meaning your device can run Muse locally or with minimal cloud dependency, depending on your build.
How to use it
- Download the open source code from Meta's repository.
- Pick a compatible display (color E Ink is confirmed) or an HDMI output board.
- Flash the firmware and configure your agent preferences.
- Connect to power and a network - Muse handles the rest.
The HDMI stick approach is particularly interesting for developers. You can plug it into any monitor or TV and have a persistent AI assistant visible at all times, without needing a dedicated screen.
Why this matters for builders
This is a shift from the usual "we ship a device, you buy it" model. By open-sourcing the code, Meta lets you control the hardware, the form factor, and the privacy boundaries. You can build a Muse gadget that only connects to your local network, or one that surfaces specific data from your own services.
For now, the builds focus on display-only interaction - showing reminders and information. But the underlying agent can handle voice and text conversations, so future hardware could include microphones and speakers.
If you've been waiting for an AI companion you can actually own and modify, this is the first major release that makes that possible without reverse-engineering a closed device.
Source: The Verge - Meta open sources code to let you make Muse AI gadgets
Source: https://www.theverge.com/tech/1004330/meta-muse-ai-gadgets-home-link
Comments
No comments yet. Start the discussion.