DIY AI pendant gains custom features before Meta’s Muse Charm launch

The Verge ·

The writer built a customised AI pendant using a $50 Waveshare touchscreen device and Meta’s Muse software, before Meta’s planned Muse Charm goes on sale. The project shows how a basic, limited AI gadget can become more personal and useful with coding help from Claude, though it takes tinkering rather than working well straight out of the box.

At first, Muse struggled to interpret commands, gave slow replies, and showed only a few words at a time; the device also had one mostly static avatar and limited camera and smart-home features. The writer used Claude Code to add a readable conversation history, spoken replies in a British voice, remote phone camera controls and previews, more avatars, and a Tamagotchi-style care game. The article presents the build as an inexpensive learning project, while noting that some features needed extra coding.

In short

  • A $50 pendant became a custom Muse AI gadget.
  • Claude Code added voice, camera tools and more avatars.
  • The project took tinkering to overcome basic limitations.

Explained simply

AI pendants are small wearable devices designed to respond to spoken requests, often using software that connects to larger AI systems. Meta’s Muse is its software for this kind of device, while the Muse Charm is a planned product that will use it.

The project pairs Muse with a low-cost touchscreen device made by Waveshare. The builder used Claude, an AI coding assistant, to add features such as saved conversations, spoken answers and camera controls; these changes required hands-on coding.

The project illustrates how people can adapt basic devices to suit their needs, but also how much work may be needed to make them more useful. For readers considering similar gadgets, it offers a view of both their potential and the tinkering they may require.

AI Culture Entertainment Technology Travel TV

Read the full article at the source →

Originally published by The Verge as “I built an AI Tamagotchi before Meta could”.

Log in to join the discussion