The platform lets makers turn ESP32 boards and Raspberry Pi systems into custom AI interfaces with displays, audio, buttons, sensors, and actuators.
Meta has introduced Muse Gadgets, a hardware platform that lets developers and electronics enthusiasts build custom physical interfaces for its Muse AI agent. The platform provides device software for ESP32 boards and Linux systems, allowing makers to connect AI functionality with displays, buttons, microphones, speakers, sensors, actuators, and other peripherals.
The ESP32 Device SDK is designed to run on off-the-shelf ESP32 development boards. Makers can connect a display to show information and images generated through Muse, add audio input and output for voice interaction, or integrate additional sensors and hardware. The platform also allows developers to adapt the software to other ESP32-based boards and create support for new hardware configurations.
For more computing capability, the Linux Device SDK can turn a Raspberry Pi or another Linux computer into a Muse gadget. This provides a route for integrating AI interaction with existing Linux applications and systems such as Home Assistant. Developers can also add their own commands and hardware interfaces to create specialised devices.
The project provides several hardware examples that demonstrate different approaches to building a Muse interface. A Raspberry Pi 5 can be used for Home Assistant and other Linux applications, while an ESP32-S3-based Waveshare board combines a 1.75-inch round AMOLED touchscreen with a speaker, microphone, push-to-talk control, and battery. Other examples include the M5Stack StickS3, an ESP32 device with a 1.9-inch colour display, and the Seeed reTerminal E1002, which uses a colour e-paper display for information such as reminders and lists.
Another proposed configuration uses an HDMI stick to connect Muse to a television, allowing information generated by the AI system to be displayed on a larger screen. The platform also lists the AiPi Lite as a compact desktop device with a button, speaker, and colour display for voice interaction and visual responses.
Meta has also developed the Muse Home Link, a small USB-C device intended to connect Muse to compatible devices on a user’s home network. It is based on Espressif’s ESP32-C5, a 32-bit RISC-V processor running at 240 MHz, and includes 8 MB of PSRAM and 8 MB of flash storage. Wireless connectivity is provided through dual-band Wi-Fi 6 supporting 2.4 GHz and 5 GHz networks. The device measures 35 × 42 × 10 mm and includes an LED for pairing and status indication.
Home Link is configured through Bluetooth Low Energy before joining the selected Wi-Fi network. Once connected, Muse can communicate with compatible devices or systems that expose a local HTTP API. Community-built integrations can extend this to devices such as lights, televisions and other home-automation hardware.
From an electronics development perspective, the platform provides a way to combine AI software with conventional embedded hardware. A simple ESP32 board can become a voice-controlled interface with a microphone, speaker and display, while a Raspberry Pi can provide a more capable Linux-based platform for automation and additional peripherals.
The device SDKs and firmware are available under the Apache 2.0 licence, with source code provided for developers to modify and use as a starting point for their own hardware projects. Meta also encourages makers to share their completed designs and repositories with the community.
The platform requires a Muse Gadget SDK token for connecting custom hardware to Muse. Meta’s current terms specify that these tokens are intended for personal, non-commercial use and place restrictions on distributing devices containing the credentials.
With support for ESP32 and Linux hardware, the platform gives electronics makers several routes for building physical AI interfaces, ranging from compact battery-powered devices and touchscreen controllers to Raspberry Pi-based automation systems and custom sensor or actuator nodes.
For more information, click here.
