At Connect this week, between the new glasses, Mark Zuckerberg held up something much smaller. A little device on a keychain with a small screen and a cartoon character on it that talks back. In his own words, it's "a joyful little device for talking to your Muse that fits on your keychain." It's called the Muse Charm, and Meta wants it shipping in December.

Most of the coverage filed it under gadget news. I think it's the most interesting thing Meta showed, and not because of the hardware. I recorded my take the day after:

Meta finally found a hardware game that works

For years Meta has been the company with the software reach and no hardware hit. The headsets found a niche and stayed there. What changed things was the Ray-Ban collaboration. Meta and EssilorLuxottica sold more than 7 million smart glasses in 2025, more than triple the year before.

The product got better, sure. But what Meta really cracked was distribution. It found a way to put its hardware in shops people already walk into, on a brand they already trust, in the hands of people who actually want it. That's the part I keep coming back to.

The Charm is a Tamagotchi for your AI assistant

Muse itself is Meta's personal assistant. It manages tasks, makes bookings, helps with work on your computer. I wrote earlier this week about what happened when Muse tried to shop on Amazon. The Charm puts that assistant on a tiny screen with an animated avatar you talk to in real time. Meta pitches it as a way to reach Muse without pulling out your phone. You can make the character your own, dress it up, play with it.

That's the clever bit. Meta is gamifying the AI assistant. A to-do app is a tool you open when you have to. A character on your keychain is something you check on because you want to. Those are completely different adoption curves, and the second one is how a product becomes a daily habit.

Same assistant, two adoption curvesA TO-DO APPA tool you openwhen you have toOpened when neededA CHARACTER ON YOUR KEYCHAINSomething you check onbecause you want toA daily habitCompletely different adoption curves.Same assistant,two adoption curvesA TO-DO APPA tool you openwhen you have toOpened when neededA CHARACTER ONYOUR KEYCHAINSomething you check onbecause you want toA daily habitCompletely differentadoption curves.
Illustrative. A tool gets opened when it has to be; a character gets checked on, and that is how a habit forms.

One device, three generations

Most AI tools today are built for people like me: mid-thirties to mid-forties, lots on our plate, happy to learn a new interface if it saves an hour. An assistant that wants to reach everyone has to work for two other groups at the same time.

It has to be simple enough for people in their sixties and older. Anyone who has designed for that audience knows how different the requirements are. Talking to a friendly face is about as low a barrier as technology gets. And it has to appeal to teenagers, who are the group Meta struggles with most, because they aren't on Facebook. A cute character they can customise is a much easier sell to them than another app from their parents' social network.

The real prize is the conversation

Here's what I think is actually going on. If people talk to this thing every day, they'll tell it what they're working on, where they want to go on holiday and what's hard in their lives right now. That's an extraordinary amount of first-party context, the kind no ad platform ever gets from clicks and likes.

For Meta that feeds everything: better products in WhatsApp and Instagram, better ad targeting, and a direct line to exactly the younger users it has been losing. It's the same logic I described in owning the customer. Whoever owns the daily conversation owns the relationship, and everything else can be sold on top of it.

How a daily conversation becomes the business1The habitYou talk to itevery day2The contextWhat you're working onHoliday plansWhat's hard right now3The payoffWhatsApp, InstagramAd targetingYounger usersWhoever owns the daily conversation owns the relationship.How a daily conversationbecomes the business1The habitYou talk to it every day2The contextWhat you're working onHoliday plansWhat's hard right now3The payoffProducts in WhatsApp, InstagramAd targetingYounger usersWhoever owns the dailyconversation owns therelationship.
Illustrative. Daily conversations produce first-party context that feeds Meta's products, ads and reach to younger users.

Whether you find that exciting or uncomfortable probably depends on how much you trust Meta with it. Either way, it's the strategy, and it's worth seeing it clearly.

Portal had the product. Ray-Ban had the channel.

Meta has tried consumer hardware outside the headsets before. Portal, the video-calling screen, was a genuinely decent product. Meta sold 600,000 units in 2020 and 800,000 in 2021, most of them to businesses, and in 2022 it stopped making consumer versions. Good product, no channel.

"New products or services may be doomed for all sorts of reasons: uneven quality, unimaginative marketing, unreliable distribution, or unrealistic pricing." Tim Brown, Change by Design

I highlighted that line long before the glasses existed, and Meta's hardware history reads like a case study of it. Portal didn't fail on quality, it failed on distribution. The glasses succeeded because Meta borrowed a distribution machine it didn't have to build itself.

Units sold, as cited in this postPortal, 2020600,000Portal, 2021800,000Smart glasses, 20257 million+PORTALGood product, no channel.Consumer versions ended 2022.RAY-BAN METABorrowed a distributionmachine it didn't build.Units sold, as citedin this postPortal, 2020600,000Portal, 2021800,000Smart glasses, 20257 million+PORTALGood product, no channel.Consumer versions ended 2022.RAY-BAN METABorrowed a distributionmachine it didn't build.
Portal units from Engadget, smart-glasses units (Meta and EssilorLuxottica) from UploadVR: same company, different channel.

That's why I'm watching the Charm. The concept is strong. What decides it is whether Meta can ramp up production and get it into enough hands by the holidays, through channels where people actually buy gifts. December isn't far off. If they pull that off, I'll be very impressed.

What I'd take from this as a marketer

The lesson isn't really about gadgets. The best product doesn't win, the product that reaches people through a channel they already use does. And the most valuable thing a brand can earn is a reason for people to come back every day without being asked.

AI makes it easy to build the product part fast. Knowing which lever actually moves adoption, the channel, the habit, the audience you're missing, is still judgement, and that's where the leverage is. Meta just gave us a very public example of getting it right after years of getting it wrong.

The Charm is a toy. The conversation it starts is the business.

Sources & further reading

External
Mark Zuckerberg on Threads: Muse Charm, shipping in December
The Gadgeteer: Meta Muse Charm wants to put your AI assistant on your keychain
UploadVR: Meta and EssilorLuxottica sold 7 million smart glasses in 2025
Engadget: Meta is discontinuing Portal devices for consumers
Tim Brown, Change by Design (Harper Business, 2009)

Related posts
Amazon blocked Meta's AI shopper. It's not about security.
Nobody needs another me-too product. Unless you already own the customer.
Most of their buyers are over 55. The brief was mobile-first.
The Death of Generic AI: Why Deep Domain Expertise is the Only Real Leverage Left

A question about this, or your own experience with it? Write to me.