San Francisco — OpenAI, Meta Test Cute AI Companion Hardware

OpenAI and Meta are each working on dedicated AI hardware built around cutesy software agents, testing whether warmth and daily usefulness can succeed where earlier AI gadgets failed, according to The Verge, which reported the effort on September 30, 2026, following OpenAI's DevDay 2026.

The push matters because dedicated AI devices have a rough track record. As The Verge's senior AI reporter Hayden Field put it in the piece, AI "has made plenty of inroads on people's phones and computers," but "it's largely failed in dedicated devices." Over the coming year, Field reported, OpenAI and Meta intend to test that assumption again, this time by pairing physical hardware with software agents designed to feel less like tools and more like pets.

What are OpenAI's 'dots' and Meta's 'Muse'?

The Verge's report centers on two in-development efforts, identified in the story's own web address as OpenAI's project "dots" and Meta's project "Muse." Both are described as consumer-facing AI agents that could eventually anchor new hardware lines, part of what Field frames as a shared strategy: use a cute, approachable software agent first, then see whether customers want a dedicated device to carry it around in.

The approach echoes the Tamagotchi, the pocket digital pet that became a 1990s fad by asking owners to feed, care for, and bond with a simple on-screen character. OpenAI and Meta appear to be betting that a similar emotional hook, layered onto modern AI, can succeed where sleeker but colder gadgets stalled.

Why has personal AI hardware failed before?

"Hardware is hard" is, in Field's words, a "tech industry mantra," and the AI sector's recent history backs that up. Field writes plainly that AI has made inroads on phones and laptops but has "largely failed in dedicated devices," without naming specific products in the portion of the story reviewed. That blunt assessment is the backdrop against which OpenAI and Meta are now making their move: both companies are entering a hardware category that has repeatedly disappointed, armed with agents meant to be likable rather than merely functional.

What's the business bet behind 'cute' AI agents?

According to The Verge, OpenAI and Meta are counting on two selling points to succeed where others failed: that the product is useful, and that it's cute. Field describes the companies as "testing appetites for physical hardware with their cutesy software agents," a sequencing that lets each company gauge demand for a companion-style AI persona in software before committing fully to a standalone device. If users warm to the character on a phone or in an app, the thinking goes, they may be more willing to pay for a physical object built to house it.

That is a notably different pitch from earlier AI hardware launches, which tended to emphasize productivity, note-taking, or voice-assistant functions over personality. The Tamagotchi framing — explicit in the Verge headline itself — signals an attempt to make the device feel like a companion worth caring for, not just a gadget worth owning.

How does this fit Meta's wider AI push?

The hardware testing described by The Verge arrives alongside other moves by Meta to monetize its AI products. An HTT News report this month detailed new AI-focused subscription plans the company has introduced, part of a broader effort to turn AI features into recurring revenue rather than one-time hardware sales. Taken together, the subscription push and the Muse hardware experiment suggest Meta is pursuing AI monetization on two tracks at once: software subscriptions for existing users, and a new physical product category built around an agent designed to be endearing.

What happens over the next year?

Field's reporting frames this as a near-term test rather than a finished product rollout, with both companies expected to move on hardware "over the next year." The Verge's account was filed from OpenAI's DevDay 2026, where Sam Altman appeared onstage as the company laid out its broader AI agenda, according to the photo caption accompanying the story. Whether either company's cute-agent strategy translates into hardware that consumers actually buy and keep — rather than another entry in the pile of AI gadgets that failed to catch on — remains the open question the next twelve months are expected to answer.

By the numbers

• Sept. 30, 2026, 6:07 PM UTC — when The Verge published its report on OpenAI's and Meta's AI hardware plans.
• 5+ years — how long Verge reporter Hayden Field has covered the AI beat, per her author bio on the story.