AI News

Muse Comes to Apple Watch: As Personal Agents Enter the Scene, AI Hardware Makers Seek Their Own Paths

LeikejiUpdated Nov 10 23:36

AI summary

Meta has announced that its Personal Agent product Muse now officially supports Apple Watch and Google Wear OS smartwatches; its long-term memory and task execution run mainly in the cloud, while the watch is only responsible for receiving instructions, reminders, and confirmations. The article argues that once personal agents are usable across devices, AI hardware lacking unique sensors and scenarios will be hit hard, while makers such as Even Realities, Plaud, and Qwen are responding with open interfaces or self-built ecosystems, respectively.

Why read this

AI score 58

Starting from Meta's integration of Muse into smartwatches, this article examines the different choices hardware makers face between openness and self-preservation once personal agents become available across devices.

Topics

Tags

Body · AI translation

Muse comes to Apple Watch: Personal Agents enter the scene, as AI hardware makers each seek their own path. Sanqi Ershiyi, 2 hours ago. The start of a new round of reshuffling in AI hardware.

Early on the morning of November 10 Beijing time, Meta announced that its Personal Agent product Muse now officially supports Apple Watch and Google Wear OS smartwatches, causing a huge stir.

Released in September this year, Muse, as a Personal Agent, can combine a user's long-term memory, personal preferences, and external services such as email and calendar to proactively identify needs and execute complex tasks. Meta had previously planned to bring it to its own AI glasses, and even launched a specially designed standalone AI hardware device, Muse Charm.

But the question that draws attention is: when Personal Agents begin to enter users' watches, earbuds, glasses, and more hardware, what changes in experience will this bring? And will it change the landscape of today's AI wearable hardware market? Muse and smartwatches are only the beginning.

But the emergence of Personal Agents gives this a new meaning. Strictly speaking, the product definition of a Personal Agent is not new; it is the ideal personal AI assistant we have heard about countless times, but the key to the explosion of Personal Agents lies in advances in AI technology.

Compared with ordinary AI assistants, Personal Agents like Muse place more emphasis on long-term memory, proactive service, and complex task execution. In interaction design, they need to continuously understand the user's life and work, maintain task continuity across different times and different applications, and proactively offer help when necessary.

This also determines that it is difficult for a Personal Agent to meet all needs relying only on a mobile app.

At this point, if a user is taking a walk outside, being able to receive a reminder and make a decision directly through a smartwatch is obviously much more convenient than taking out a phone and reopening an app.

Moreover, a Personal Agent does not need to run on the watch all the time. Muse's long-term memory and task execution mainly happen in the cloud, and the watch can be responsible only for receiving instructions, reminders, and confirmations. This also lowers the technical threshold for Personal Agents to enter different hardware devices.

Meta certainly understands this. At the Connect conference in September, Meta announced that Muse would connect to AI glasses, and in the future users may even be able to directly let Muse understand what they see before their eyes and then execute corresponding tasks. Muse Charm, which appeared at the same event, aims to let users carry a dedicated Muse entry point with them.

Muse Charm, image source: Meta

From phones and computers to smart glasses, smartwatches, and then standalone hardware, Meta is expanding Muse's usage scenarios as much as possible.

Meta will not be the only company on this path.

China's Qwen GO! has already begun internal testing, trying to use the Alibaba ecosystem to complete personal matters such as shopping, travel, and appointments; Today is likewise building its own Personal Agent around long-term memory, proactive reminders, and cross-application task execution.

Although there is not yet sufficient evidence to show that they have already landed on mainstream smartwatches like Muse, this can also be said to be a matter of course. After all, if a Personal Agent wants to continuously serve users, it needs to integrate into users' daily lives as much as possible. And smartwatches, glasses, and earbuds beyond phones and computers happen to provide more opportunities to reach users.

More importantly, the more a Personal Agent can be used across devices, the less users need to specially buy a new piece of hardware just to obtain a certain AI capability. Personal Agents drop in: welcome or vigilance?

The good side is straightforward. Many AI hardware products today still remain at functions such as voice Q&A, recording summaries, and simple reminders. Introducing mature Personal Agents gives these devices a chance to gain long-term memory, proactive service, and complex task execution capabilities, and even to expand into usage scenarios they could not originally cover.

The problem is that when devices from different brands can all call the same Personal Agent, the AI features originally provided by manufacturers themselves may also gradually lose appeal.

This impact may be even greater for products that lack unique sensors, interaction capabilities, and usage scenarios, and merely put an AI assistant into a new shell. Consumers already have phones, watches, and earbuds. If existing devices can conveniently use Muse, why would they buy an AI necklace or AI brooch with similar functions?

Of course, they do not necessarily have to respond passively.

In fact, some manufacturers have already begun trying to open up their own AI capabilities and hardware interfaces. Although what they connect to is mainly ordinary Agents, tools, or external AI services, which cannot be directly equated with supporting Personal Agents, these practices can help us understand the choices of different manufacturers.

For example, smart glasses manufacturer Even Realities. Even G2 supports connecting to external tools such as OpenClaw through custom Agent configurations, and can also interact with AI programming tools on computers through Terminal Mode.

Even G2 has no camera or speaker, relying mainly on a lightweight display, voice input, and a companion ring for interaction. If Muse and Today could one day use Even G2 to provide continuous task reminders, progress feedback, and action confirmations, that could give users a new reason to buy.

Even G2, image source: Even Realities

For Even Realities, rather than investing a large amount of resources in developing a Personal Agent capable of competing with Meta and Google, it may be a more realistic choice to let more mature AI products deliver a good experience on its glasses.

There is also Plaud.

Earlier this month, Plaud officially launched MCP and CLI, allowing third-party AI tools such as ChatGPT and Claude to read user-authorized recordings, transcripts, and summaries. Users can hand these directly to AI tools they are more familiar with and continue using them alongside other work materials.

Plaud Note Pro, image source: Plaud

For Plaud, this can expand the value of its recording devices. As long as users still need reliable recording, transcription, and professional information capture capabilities, they still have a reason to use Plaud. In fact, QwenNote A2 has a similar design: by connecting to computer Agents such as Codex and Claude Code via CLI, Codex/ChatGPT Work can directly use QwenNote A2's transcripts or summaries.

Interestingly, however, Plaud also launched its own Plaud Agent during the same period, beginning to expand into cross-application workflows, document generation, and task execution. Opening up data while strengthening its own AI services precisely shows that Plaud also knows it cannot leave all subsequent software value to third parties.

If Muse or Today could one day directly read Plaud recordings and combine them with the user's emails, calendar, and work materials to complete follow-up tasks, whether users would still need to pay for some of Plaud's own advanced AI services becomes a real question.

By contrast, the situation with Qwen is different. At the recent Yunqi Conference, Qwen launched in one go a new generation of Qwen AI glasses, the N1 series, the QwenNote A2 recording card, the QwenNote Eva desktop robot, and even hardware products such as the Qwen computer, along with its own Personal Agent, Qwen GO.

For Qwen, which has models, applications, a service ecosystem, and hardware, it has reason to let third parties provide more specialized capabilities while continuing to use Qwen GO! to help users handle personal affairs and meet various service needs.

For manufacturers like Even Realities, openness may make hardware more attractive; Plaud needs to find a balance between professional data services and its own AI business; while manufacturers like Qwen and Meta that also have complete AI ecosystems have a stronger incentive to retain the main Personal Agent entry point.

Ultimately, the value of hardware itself will not disappear just because Personal Agents become more powerful. Professional recording devices still need reliable audio capture capabilities, AI glasses need sufficiently good display, camera, and wearing experience, and smartwatches cannot do without long-term health monitoring, sports algorithms, and mature interaction systems.

To be fair, Muse's arrival on Apple Watch may not immediately have much impact on the smartwatch market. Apple and Google already have their own AI assistants, and Muse is still far from obtaining full system-level permissions.

But on the one hand, Muse can already be used through ordinary smartwatches, and there are also plans to launch Muse Charm, hoping to give users a standalone device for using Muse on the go. For users, they may no longer need to get to know a new AI assistant and rebuild their personal information on every piece of AI hardware.

Instead, they would rather have the Personal Agent they use over the long term continue naturally across different devices.

On the other hand, for hardware manufacturers, they can use third-party Personal Agents to expand product value, but they may also lose part of the AI services and user relationships they originally controlled.

Image source for this article: 123RF licensed image library Infringement report Return to sharing Related articles All comments

Source

Leikeji