Avi Schiffmann announced "friend 2.0" on July 30, 2026, in a post on X, introducing an updated version of the Friend AI wearable that adds a built-in speaker capable of projecting a consistent voice personality to the wearer (TechCrunch).

The original Friend pendant launched in July 2024 as a text-message-based AI companion positioned to combat loneliness. The 2.0 overhaul shifts the interaction model from a purely text-based exchange to one where the device speaks back. A new commercial published on YouTube demonstrates the voice feature in action.

The retail price moves to $249, a substantial increase from the original launch price. TechCrunch cites the first-generation launch price as $99, while Wired reports the original pendant was priced at $129 (Wired). The Wired figure may reflect a post-launch price adjustment or a different configuration; regardless of which baseline applies, the jump to $249 represents a significant increase over either cited figure.

Schiffmann paid $1.8 million for the Friend.com domain, according to Wired. That purchase underscores the brand positioning he is pursuing, which he articulated in a separate X post with unconventional framing: "not an assistant, and it is not a lover" but rather some kind of "confidant, friend, God." The language is deliberately outside the standard product categories that dominate the AI hardware landscape, where the default framing is assistant, copilot, or productivity tool.

The pricing and positioning taken together invite scrutiny. At $249, Friend 2.0 enters a price band occupied by capable smartwatches and mid-range earbuds, devices that leverage mature hardware ecosystems and broad third-party software support. The value proposition here depends almost entirely on the quality and persistence of the companion relationship the device can sustain, rather than on functional utility metrics like task completion rate or integration depth.

The addition of a voice personality is the most consequential hardware change. Text-based AI companionship has existed for years across mobile apps, and the original Friend differentiated itself primarily through form factor, a wearable pendant rather than a phone screen. A built-in speaker that projects a consistent voice moves the device closer to the category of ambient AI, where interaction is hands-free and always-available rather than requiring the user to look at a screen. The consistency of that voice personality across sessions is the technical detail that determines whether the device feels like a companion or a novelty. If the voice, tone, and conversational memory drift between interactions, the illusion of relationship fractures quickly. If it holds, the experience could be meaningfully different from what a phone-based chatbot delivers.

The broader question for AI wearable hardware is whether a dedicated single-purpose device can sustain consumer interest in a market where general-purpose smartphones already run increasingly capable LLM-powered assistants. The track record for standalone AI hardware is thin. Devices that promised ambient intelligence have struggled to justify their existence alongside the phone most users already carry. Friend 2.0's differentiation hinges on emotional positioning rather than functional capability, and Schiffmann's language about "confidant" and "God" makes that intent explicit.

Worth noting is the $1.8 million domain investment alongside the deliberate decision to frame the product outside conventional categories. Whether the emotional-companion angle sustains a $249 hardware business in a category where functional-assistant framing has so far failed to generate sticky adoption remains the open question for this product cycle.

Friend has released a promotional video on YouTube demonstrating the voice feature, giving potential buyers a concrete look at the interaction model before purchase.