Situation report: The AI hardware market has been a graveyard so far. Two of the biggest players in AI think they’ve found a way through it. According to The Verge AI, Meta and OpenAI are both preparing dedicated AI devices over the next year. They’re using the same playbook: get people attached to cute software agents first, then sell them the hardware.
That’s an opportunity and a risk at once. Here’s the intel.
🎯 What Happened
- Meta’s Muse Charm. Mark Zuckerberg compared it to a keychain. It’s a pendant-like device with a screen that shows Meta’s Muse AI agent, a cuddly character that gives the whole thing a Tamagotchi feel. Meta wants it out before this year’s holiday shopping season.
- OpenAI’s Jony Ive device. OpenAI is building hardware with the famed ex-Apple designer. A public filing says it won’t ship before February 2027.
- OpenAI’s Dots platform. The hardware didn’t show up at DevDay. What did show up was Dots, a major new AI agent platform with personalizable colored blobs that have eyes.
Asked whether Dots would end up in a physical device, Sam Altman said that “seems like a very reasonable thing to assume we may do someday.” He wouldn’t say more.
Zuckerberg was more direct about the Charm. “We’ve packed a lot of technology into this little guy,” he said onstage. “So, if you’re not wearing glasses, this is going to be by far the fastest way to talk to your Muse and to show it what’s going on around you.”
🧭 The Strategy: Software First, Hardware Second
“Hardware is hard” is an old line in tech, and recent AI gadgets keep proving it. The Humane AI Pin and the Friend pendant mostly drew frustration and backlash. With public anger toward AI growing, the next launch faces even worse conditions.
So Meta and OpenAI are taking a different route:
- Soft-launch the agent. Get people using it on the phones and computers they already own.
- Build attachment. Cute, personalized characters give people a reason to care about the agent, not just use it.
- Let word of mouth do the work. The hope is that users start telling friends that AI agents “really work now.”
- Then ship the body. Once people value the agent, a physical version feels like an upgrade, not a gamble.
What stands out to me is how much this flips the Humane approach. Humane asked people to buy a $700 device for an assistant they’d never used. Meta and OpenAI want the relationship to exist before you ever see a price tag.
Friend tried a similar software-first approach. But Friend sold a companion chatbot. Meta and OpenAI are selling productivity tools, and that’s a much easier pitch to justify.
🔍 Where OpenAI Is Taking ChatGPT
OpenAI’s comments point toward a future where the interface stops mattering much. Thibault Sottiaux, OpenAI’s head of core products, said the company would be “evolving ChatGPT in place to be this incredible personal assistant very soon.”
He expects that as people reach their agents by email, text, or phone call, the point of interaction is “all going to blur at some point.” It’ll be a “natural part of your life,” he said.
Altman’s bet is that Dots handles long-running work better than earlier agents did. “This feels like an always-on helper, collaborator … that you can give very large pieces of work to and it gets to know you and helps proactively over time,” he said.
⚠️ Threat Assessment
The demos haven’t been clean, and they need to be.
- Muse has already caused trouble. The Verge AI points to its “concerning exploits” in Facebook Marketplace.
- Dots stumbled live. A voice demo at DevDay went awry onstage.
- Trust is thin. After years of underwhelming agents, people won’t give a new device many chances.
- Backlash is real. Wearable AI that watches “what’s going on around you” will get hard questions about privacy.
📋 What This Means for You
If you build on AI platforms, pay attention now. Both companies are pushing agents that run across many channels and handle long tasks on their own. That changes what an “app” looks like and where your users meet your product.
If you’re a buyer, don’t hurry. Meta’s Charm lands first, so it’ll be the real-world test of whether the software-first strategy works.
The holiday season will give us the first verdict. The full story is at The Verge AI.