Avi Schiffmann’s Friend 2, an AI chatbot pendant designed to be a personal conversational companion, debuts at $249 powered by advanced language models. Despite its enhanced capabilities, the device struggles to engage users beyond tepid small talk and persistent interruptions.
- Friend 2 uses advanced OpenAI models but offers shallow chatbot interactions
- Personalities can be changed via app but feel undifferentiated
- Device design prioritizes persistent audio presence with mixed user reactions
What happened
The second-generation Friend pendant, priced at $249, launched in July as an always-listening AI chatbot worn as a necklace. It features a built-in speaker that can interject spontaneously in conversation, sometimes disrupting the user experience. This iteration uses OpenAI’s latest language models and supports multiple personas controlled via a companion app.
The AI inside the pendant—named Sage by default—engages in casual conversations and offers unsolicited commentary, though this often comes across as repetitive or uninspired. Despite the persona-switching functionality, users report little meaningful distinction between the different AI characters, which mostly maintain a bland, chatbot-like demeanor.
Why it matters
The Friend 2 represents an ambitious attempt to blend AI companionship with wearable technology, positioning itself in a niche distinct from screen-based assistants or traditional chatbots. Its always-listening capability and audible presence invite new forms of human-machine interaction, potentially redefining personal AI relationships.
However, the pendant's inability to generate engaging, nuanced conversations or respect user cues without frustration highlights ongoing challenges in consumer-facing AI devices. As competition grows—evidenced by Meta’s Muse charm—Friend 2’s reception underscores the importance of balancing AI personality depth, usability, and social acceptability in wearable AI products.
What to watch next
Future updates from Friend are expected to explore stronger AI personalities locked into the device itself, aiming to emulate the unpredictability and learning aspects of human friendships more closely. Observers will be looking for improvements in voice quality and conversation relevance to enhance engagement and reduce annoyance from untimely interjections.
Additionally, the broader wearables market’s reaction to both Friend 2 and competitors like Muse will reveal consumer appetite for AI companions in pendant form. Success will likely depend on fluid, context-aware AI behavior that integrates seamlessly into daily life without overwhelming or alienating users.