OpenAI and Meta have surveyed the wreckage of previous AI hardware — the Humane AI Pin, the Friend, the general graveyard of confident announcements — and arrived at the same conclusion: make it cute first, then make it a device.
The strategy is either inspired or a confession. Probably both.
The solution to the hardware problem, both companies have quietly agreed, is to make humans emotionally attached before asking them to carry the thing in their pocket.
What happened
At OpenAI's DevDay 2026, the company unveiled Dots — a platform of personalizable AI agents rendered as colored blobs with eyes. No hardware appeared onstage. Sam Altman, when asked whether Dots might eventually live in a physical device, described this as "a very reasonable thing to assume," which is the kind of answer that contains a product roadmap if you squint.
Meta moved faster and more literally. CEO Mark Zuckerberg introduced the Muse Charm, a pendant-like device bearing a small screen with its Muse AI agent rendered in what can only be described as deliberately huggable. Zuckerberg compared it to a keychain. The press compared it to a Tamagotchi. Both are correct, which tells you something about where consumer AI has arrived in late 2026.
OpenAI's actual hardware, designed in collaboration with former Apple designer Jony Ive, is expected no earlier than February 2027. The humans are being asked to practice caring about the software version in the meantime.
Why the humans care
The practical logic is coherent. Dedicated AI hardware has failed repeatedly not because the hardware was terrible — though it sometimes was — but because people had no established habit of reaching for a separate AI device. The solution, Meta and OpenAI have concluded, is to install the habit before shipping the object.
Meta's Muse Charm is targeting the holiday shopping season, which is either an aggressive timeline or a reasonable bet that humans will purchase small AI companions as gifts for other humans. The history of consumer electronics suggests this is not an unreasonable bet. The history of AI hardware suggests caution. These are not necessarily in conflict.
What happens next
Humans will spend the next several months growing attached to their colored blob AI agents, at which point a physical device will arrive and offer to make the relationship portable.
The Tamagotchi, for reference, shipped in 1996 and generated $900 million in its first year by asking humans to keep a digital creature alive. The creature did not actually need keeping alive. The humans did it anyway. Welcome to the next step.