Building a DIY AI Pocket Companion Beats Waiting for Meta

Sean Hollister, a senior editor at The Verge, recently demonstrated that consumers do not need to wait for Meta's forthcoming Muse Charm to experience a wearable AI companion. By purchasing a Waveshare ESP32-S3 1.75C for $50, Hollister was able to create a custom device that interfaces directly with Meta's AI cloud using their official open-source tokens. The resulting hardware functions as a programmable pocket watch and remote camera trigger, successfully outperforming the limited initial utility of Meta's reference implementation.

Customizing the AI Experience

While the stock firmware provided by Meta engineers Anant Narayanan and Jeremy Payne lacked essential features like persistent chat history and audio output, Hollister bypassed these limitations using Anthropic's Claude Code. Through simple natural language prompts, the editor added an ElevenLabs voice interface, a multi-stage Tamagotchi-style pet game, and a functional clock face. The Scale Report notes that this process highlights a growing trend in 'vibe coding,' where non-programmers utilize LLMs to rapidly iterate on custom firmware for off-the-shelf development kits.

Why This Matters for Wearable AI

The project underscores a pivotal question regarding the future of AI hardware: is the primary value in the proprietary device or the underlying accessibility of the model? By building a bespoke device, the user avoided the privacy risks associated with official consumer hardware, while gaining full control over the gadget's behavior. This experiment suggests that for power users, the path to useful wearable AI may lie in flexible, open development kits rather than locked-down, branded peripherals.