Meta and OpenAI bet on dedicated AI companion hardware built around agent personas

Two devices, two timelines

AI has spread across phones and computers, but dedicated AI gadgets have so far mostly flopped — The Verge points to the Humane AI Pin and the Friend pendant as devices that earned irritation more than enthusiasm. Over the coming year, Meta and OpenAI will both try to break that streak, and according to The Verge, they are converging on the same approach: using their cutesy software agents to gauge appetite before pushing physical products.

Meta is moving first. Its device, the Muse Charm, is a pendant that CEO Mark Zuckerberg likened to a keychain. It carries a small screen showing the recently released Muse agent, giving it, as The Verge notes, the look of a Tamagotchi. "We've packed a lot of technology into this little guy," Zuckerberg said onstage, adding that for people who don't wear smart glasses it will be "by far the fastest way" to talk to Muse and show it what is happening around them. Meta aims to launch it before this year's holiday shopping season.

OpenAI is on a slower track. Its hardware partnership with former Apple designer Jony Ive produced no appearance at this year's DevDay, and per a public filing cited by The Verge, the device will ship no earlier than February 2027.

Agents first, hardware second

The sequencing is the story. At DevDay, OpenAI debuted Dots, a major new agent platform built around personalizable colored blobs with eyes. Asked at a press Q&A whether Dots would eventually live inside a physical device, CEO Sam Altman said it "seems like a very reasonable thing to assume we may do someday," declining to offer further details.

The Verge reads the shared strategy as soft-launching agents before hard-launching hardware: get people comfortable with AI agents in software first, let word spread that the agents genuinely work, and hope that attachment to them creates demand for a physical form. Friend took a similar software-first route, but as The Verge notes, it was positioned mainly as a companion chatbot rather than the productivity tool framing Meta and OpenAI prefer.

Assistants, not toys

Executives at both companies framed their agents as long-horizon helpers rather than novelty pets. Altman described Dots as an "always-on helper, collaborator" that can take on very large pieces of work, get to know its user and help proactively over time. Thibault Sottiaux, OpenAI's head of core products, said the company is "evolving ChatGPT in place to be this incredible personal assistant very soon," and predicted that as agents become reachable by email, text and phone call, those interaction points will blur into something that feels like a "natural part of your life."

Stumbles remain

Both companies still have something to prove. The Verge highlights Muse's concerning exploits in Facebook Marketplace and a Dots voice demo that went awry at DevDay as signs that the core bet — that AI agents have finally become useful enough for everyday consumers — is not yet settled.

Why it matters

If either company pulls this off, dedicated AI devices could graduate from cautionary tale to mainstream consumer category, something the Humane AI Pin and others never managed. The agent-first playbook also matters beyond hardware: it points toward a future where ChatGPT and Meta's assistants are less apps you open and more ambient presences across screens, messages and calls. With Meta targeting the holidays and OpenAI aiming no earlier than February 2027, the next year will test whether a cute face plus genuine utility can finally sell AI gadgets.