Meta’s pitch for AI hardware just became more concrete. At Connect 2026, Meta introduced Muse Charm, a pocket device built around Muse, the company’s personal AI agent.
Meta said Muse is coming to its AI glasses in the coming months, with the agent able to act on what the wearer is looking at. The company also announced new Muse connectors for shopping, payments, travel, work tools and an email address that the agent can use to communicate or complete tasks.
Muse Charm is the more unusual piece. Meta’s official Connect post described it as a pocket device for talking to and interacting with Muse, powered by a real-time voice model. Meta said it would share more details later this year.
Early event coverage filled in more of the shape. Axios described Muse Charm as a keychain-sized handheld device, while TechCrunch reported that Mark Zuckerberg showed a small screen for the Muse avatar and a fingerprint sensor used to start talking without unlocking a phone. The Independent reported that the device has a roughly 2-inch touchscreen and built-in 5G connection, and that Meta did not disclose a price.
No Canadian pricing or availability for Muse Charm has been announced in the official materials reviewed for this article.
Why Charm is different from the last wave of AI gadgets
Most dedicated AI hardware has faced the same problem: the gadget arrives before the reason to carry it.
Meta may be approaching the market from the other direction. Muse launched Sept. 8 in the U.S. on iOS, Android and the web. Meta describes it as a personal AI agent that can open a browser, fill out forms, negotiate, keep working after the app is closed and return when approval is needed for actions such as sending an email or making a purchase.
A few days later, Axios reported that Muse had reached No. 1 among free iPhone apps in the U.S. App charts are an early signal, not proof of long-term use. Still, they give Meta something many AI hardware startups lacked at launch: an agent people are already trying before the dedicated device ships.
That changes the hardware pitch. Charm is not trying to replace the phone on day one. It is trying to make Muse faster to access, easier to personalize and harder to ignore.
The hard part is permission, not packaging
The more useful a personal agent becomes, the more access it needs. According to Meta’s Muse launch materials, the agent can connect to services, remember user-provided details, work through a dedicated cloud VM and ask for approval before sensitive actions. Meta also says Muse does not share conversations or VM data with Meta’s ad systems and that people can opt out of having interactions used to train Meta’s AI models.
Meta’s security argument now extends into wearables. In a separate engineering post, the company described Private Processing for AI glasses, a confidential computing system designed so that AI workloads can run in cloud infrastructure without Meta being able to access the processed data. Meta said it is expanding bug bounty coverage and working with outside security researchers to audit the approach.
Those details matter because Charm sits in a more intimate category than a chatbot. A pocket AI companion, glasses-based agent or voice-first device is not just software. It becomes a permission system people carry with them.
That is where the consumer test gets tougher. Convenience has to outweigh the unease of giving an agent access to accounts, payments, location, voice, visual context and memory.
AI hardware has a rough recent history
The caution around Charm is not theoretical.
Humane’s AI Pin was one of the most visible attempts to build a post-smartphone AI device. In February 2025, HP announced plans to acquire key AI capabilities from Humane, including its Cosmos platform, talent and intellectual property. TechCrunch reported the same day that Humane was discontinuing AI Pin sales and that existing devices would lose cloud-based functions after Feb. 28, 2025.
Rabbit’s R1 faced a related problem. WIRED’s review argued that the R1 did not offer enough beyond a smartphone to justify carrying another device, while Engadget wrote that phones could already handle similar AI tasks more easily.
Those examples do not prove Charm will fail. They show the hurdle. A dedicated AI device has to be more than charming hardware. It has to remove friction the phone still creates.
Hardware will not solve the platform problem
Muse also faces a challenge that no keychain can fix on its own: other platforms have to decide whether they want an outside agent acting on their services.
Axios reported that Amazon blocked Muse from browsing and buying products on its platform less than two weeks after Muse launched. Amazon cited concerns around customer security, user experience and authorization for third-party agents making purchases on behalf of customers.
Meta has announced a growing list of retail and work connectors, including Shopify catalogue access, Walmart, Best Buy, Sephora, Wayfair, Notion, GitHub and Box. But agent usefulness depends on coverage. If major services resist external agents, dedicated hardware can feel less powerful, no matter how polished it looks.
The details that will decide Charm’s chances
Charm is still early enough that the missing details are as important as the announcement. The device’s prospects will depend on a few practical questions.
- Price: Meta has not announced what Charm will cost.
- Plans and connectivity: early reports describe built-in 5G, but service costs and carrier details have not been laid out.
- Battery life: a pocket AI device has to last long enough to justify carrying it separately.
- Availability: Meta has not announced Canadian availability for Charm.
- Permissions: Muse has to make approval flows safe without making them annoying.
- Use case: Charm needs one repeatable task that feels faster than using a phone, watch, earbuds or glasses.
Meta’s advantage is distribution. The company can push Muse through apps, WhatsApp, glasses and, now, hardware. OpenAI has also moved toward hardware by merging Jony Ive’s io Products team into OpenAI, while other companies continue to search for an AI-native device category.
Charm gives Meta a visible symbol for that race. It does not settle the race.
The company now has a consumer AI agent with early momentum and a hardware strategy built around it. Turning that agent into a device people actually want to carry is the next test. If Charm makes Muse faster, more useful and easier to trust, it could make dedicated AI hardware feel less speculative. If it does not, it risks becoming another reminder that the best AI device may still be the one already in the pocket.

Tech Help Canada Staff researches, writes, and reviews practical content for business owners and professionals. Our coverage spans business, marketing, SEO, technology, and the tools and systems people use to grow and operate online. We focus on clear, useful information backed by research, hands-on experience, and editorial review. Learn more about our team and editorial standards. Need help with something? Contact Us







