过去四个月,笔者曾先后感受过两款「AI 耳机」:光帆 AI 全感穿戴设备和韶音 OpenFit 2 AI。
前者给耳机装上摄像头,配上智能手表和能独立联网的耳机盒,试图让 AI 跟着人一起看、一起听;后者则沿着韶音熟悉的开放式耳机路线,接入录音、转写、总结和翻译能力,摇身一变成为了工作情形下的 AI 生产力助手。

两款机型都可以称为 AI 耳机,应用情形却保持着一些距离。
关联阅读:带摄像头的 AI 耳机能替代 iPhone 吗?我戴了 72 小时之后有了答案
关联阅读:韶音 OpenFit 2 AI 感受:如何用 AI 改造耳机?
但如果,它们来一次「合体」呢?
就在最近,韶音在新品耳机中引入了光帆自研 AI OS 平台。从此,这副耳机在录音、转写之外,还可以完成创建日程、备忘提醒、总结工作消息等任务。
这是一场很有意思的合作:作为音频机型专家的韶音,懂得怎么把耳机做得动听也好戴;而光帆则在探索更面向将来的事情:长时间伴随在笔者耳边的 AI,究竟能怎样真正地帮到笔者。

当这两种思路落到同一副耳机上,变化也就不只是多了几个 AI 能力,而是 AI 耳机开始从「记录信息」,进一步走向对全天候上下文的「深度处理」。
感受完这次耳机升级,笔者发现一个趋势已然形成:AI 硬件的竞争,正在从单点能力走向平台能力。
单一能力很难持续满足使用者不断增长的需要,多能力协同才是 AI 时代更关键的方向。借助 AI OS 统一调用多元能力,让硬件不再是单一能力的载体,而成为连接生活与工作的智能入口,最终才能形成真正有效的情形闭环。
会议前后,AI 还能做更多
先回到韶音 OpenFit 2 AI 原来的长处。
在此前的感受中,笔者对它印象最深的一点,是它把「录音」这件事做得足够丝滑。戴着耳机,可以记录通话或临时交流;放进充电盒,又能摆在桌上收录会议。随后,声音交由软件转写、总结,整理为一份便于回看的文字资料。

这是一条过去一年里被许多「AI 耳机」重复验证过的成熟流程。它贴近耳机天然擅长的声音情形,也为 AI 切入工作情形找到了一个很好的入口。
但「办一件事」所需要的,往往不只是转写。
什么时候开会、会后要给谁发资料、哪天交初稿、下一次沟通在何时何地,这些内容即使经由文字处理,最终还是要写入日历、to-do 和备忘录,才能用于推进后续工作。
更关键的是,转写后的资料是「固定」的,而生活和工作向来充满「流动性」。

作为编辑,我的日程并不总会按照最初的安排开展。一个临时的需要,可能会打乱原有的时间线;走路去吃饭的路上,也可能随时想到需要补充的内容。
如果用的是「原版」的 OpenFit 2 AI,我这时大概只有两个挑选:要么另外录音口述新的安排和念头,过后再重新整理;要么当即掏出手机,修改备忘录和日历。
光帆此次带来的能力,意义恰好在于:第一时间在口头上处理信息。

「智能消息」过滤推送不断更新的沟通内容,「日程待办」承接时间与任务,「随口记」保存随时冒出来的想法……这意味着,在我走路、开车乃至做饭期间,我都能随时给 AI 发出指令,「动动嘴」就把事务安排得井井有条。
说给耳机听,也交给耳机办
用语音来控制工作流的逻辑,和光帆 AI 全感穿戴设备本身的机型形态关系密不可分——本质上,它的智能中枢是那台配备了 eSIM 和 GPS 的耳机盒,而耳机盒并没有 GUI 界面可言。

因而,从外观之初,它就优先用语音来处理工作。
即使这一平台跳出自家硬件,「移植」到了以手机为智能中枢的 OpenFit 2 AI 之上,它的基本逻辑仍旧没有变化。对于不想时常掏手机的我来说,语音交互显然是一种很有用的能力升级。
最容易理解的例子,就是安排一场会议。
我现在可以直接说,「晓帆,帮我记一下,明天下午 3 点到隔壁楼开沟通会」,此时 AI 会主动为我记下日程、设置提醒——只要第二天耳机还挂在耳边,会议前夕它就会发起语音提醒,起到一个全天候督促的作用。
当然,如果在那之前我已经给同一时段安排了别的活动,AI 一开始便会提醒存在日程冲突,建议我更改时间;一旦会议取消,用语音删除日程也是一句话的事。

同一件事,从建立、查询到修改,都沿着对话继续开展。这个全天候可调用的日程助手,显然比点按手机直观得多。
备忘录也有类似的变化。
每个人大概都有过这样的经历:想到一个绝妙的想法,但拿起手机准备记录时却被一条消息吸引注意力——等看完消息,刚刚浮现的想法就已经烟消云散了。
对于这种需要「数字排毒」的情形,光帆给出的交互方式是,说一句「晓帆,帮我记个灵感」,口述即生成一条备忘,全程不被其他信息打断;过后,也能继续用语音查询或修改这条备忘。甚至在我记不太清原话时,照样能找出记录。

此前在感受光帆 AI 耳机时笔者就注意到,它很合适充当信息漏斗,为我筛选出微信、飞书消息中的有用信息。如今,同样的能力也跟随光帆 AI OS 来到了韶音耳机之上——
当消息不断涌入,AI 会自动过滤掉闲聊,只将最关键的信息以语音形式推送到耳边。不同使用者可以依据自己的习惯来挑选「快速」、「中速」或「慢速」总结方式,来应对不同的工作和社交情形。
上班路上或收拾行李,照样能从容跟进联系人和工作群里不断跳出的动向。

一系列感受下来,笔者可以留意到,光帆这套平台化 AI 思路的核心卖点,实际上是一种软件感受。
如果说录音负责的是记录信息,那光帆 AI OS 负责的就是调用、处理这些信息,在它的软件服务里为笔者搭建出一套工作流。
既然是一种软件生态,它需要的东西实际上很简单:输入信息的麦克风(或许还有摄像头和 GPS 芯片)、将信息上传到云端交由 AI 处理的信息通路,还有最终用于回馈信息的扬声器。

而这几种必需的硬件,早已存在于 AI 耳机、智能眼镜和其他智能配饰上。
也就是说,光帆 AI 不必绑定在光帆自家的硬件中。它完全可以作为一种 AI 服务应对思路,进驻其他品牌的硬件设备。
引入现成 AI,硬件有了另一种做法
《庄子》里有这样一个故事:某个宋人家族善于制作防止手皴裂的药,世代用于漂洗丝絮;后来,一位客人买下药方,将它用于冬日水战,协助吴王取胜,最后得到封赏。
工艺常有类似的际遇。诞生时,它服务于一个具体的难题;但从最初的机型扩散开来,仍会发掘出无限多的新用途、新可能。

光帆的 AI 服务能力,也可以是这样的一张「药方」。
在自家的全感穿戴设备上,光帆 AI 需要配合耳机、耳机盒和手表工作,探索一套相对完整的随身 AI 感受。而进入韶音 OpenFit 2 AI 时,它舍弃了更「重」的视觉识别和独立联网能力,却和韶音原有的「录音-转写-总结」能力发生了奇妙的化学反应。
大量录音素材为日程、备忘和消息处理给出了多元的上下文,而光帆 AI 也在这副原来就合适长时间佩戴的开放式耳机上,达成了始终及时的消息播报。
对韶音的原有使用者来说,他们可以继续用熟悉的耳机听歌、通话、录音,同时尝试把几件过去需要掏出手机处理的事情交由新来的 AI 工作流。硬件原有的长处,也因而有了更多发挥空间。

这恰好回应了许多硬件品牌面对的难题:机型已经做得不错,下一步还能给使用者什么?
音质、舒适度、电池续航当然值得继续打磨,但这些进步通常发生在既有用途之内。AI 则给出了增加用途的可能。一副耳机能否帮忙管理安排,一副眼镜能否随口记下见闻,都关系到使用者会在多少个时刻想起它、使用它。
硬件交付之后,软件仍然可以更新,使用情形也可以继续增加。对于品牌而言,这也是在机型交付之后赋予机型更多价值,延长机型与使用者关系的关键方式。

过去阻碍为硬件增加新能力的一大因素在于研发成本——科技硬件品牌很多,但真正具备庞大终端规模,足以将软件投入分摊到大量设备上的公司却屈指可数。
对一家专注于特定品类的硬件厂商来说,即使机型卖得不错,也不意味着它合适独自承担模型调度、语音交互、服务接入和长期维护的全部工作。
就此而言,光帆给出的现成应对思路相当有吸引力。
使用外部思路,也就多了一种分配研发资源的方式。韶音可以继续打磨擅长的声学、佩戴和录音感受,运动耳机品牌也可以在传感器、结构和算法上继续精进,同时借助光帆在信息处理上的积累,为机型增加新的用途。

这实际上是消费电子行业一贯的习惯。
以常见的 TWS 耳机为例,降噪、音效和连接工艺背后,都有面向多家厂商的思路给出商。品牌在这些共同基础上,继续做自己的声学外观、交互和机型取舍。使用外部工艺,并不妨碍最终机型各有性格。
光帆 AI 给出的,正是这样一种让专业能力在品牌之间流动的模式。
它将围绕语音交互组织能力的思路,以及连接对话、管理信息和执行操作的经验带入合作中,硬件品牌则可以以此为基础,挑选合适自己的能力,继续定义机型。

在韶音这里,结果就是开放式佩戴与随身事务处理的结合。将来如果进入其他机型,同一套工艺底座还可能依据使用条件重新组合。合作品牌越熟悉自己的使用者,越有可能挖掘出 AI 能力的新用途。
至于这套能力还能做什么,将来就要交给市面来验证了。
本文来源:https://www.ifanr.com/1681475
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。
Over the past four months, I have had the chance to experience two "AI earphones": the LightSail AI full-sense wearable device and the Shokz OpenFit 2 AI.
The former equips the earphones with a camera, pairs them with a smartwatch and a charging case that can connect to the internet independently, attempting to let AI see and hear alongside the user; the latter follows Shokz's familiar open-ear design route and integrates recording, transcription, summarization, and translation capabilities, transforming into an AI productivity assistant for work scenarios.
Both models can be called AI earphones, yet their use cases remain quite distinct.
Related reading: Can AI earphones with cameras replace the iPhone? I got my answer after wearing them for 72 hours.
Related reading: Shokz OpenFit 2 AI review: How to use AI to transform earphones?
But what if they "merge"?
Recently, Shokz introduced LightSail's self-developed AI OS platform into its new earphones. From now on, beyond recording and transcription, these earphones can also create calendar events, set reminders, and summarize work messages.
This is a fascinating collaboration: Shokz, as an audio device expert, knows how to make earphones that sound great and feel comfortable; meanwhile, LightSail is exploring something more future-oriented: how an AI that stays by my ear all day can genuinely help me.
When these two approaches land on the same pair of earphones, the change is not just about adding a few AI features; it's about AI earphones moving from "recording information" toward "deep processing of all-day context."
After experiencing this earphone upgrade, I noticed a trend taking shape: competition in AI hardware is shifting from single-point capabilities to platform capabilities.
Single capabilities are increasingly unable to sustain the growing needs of users; multi-capability collaboration is the more critical direction for the AI era. By leveraging an AI OS to orchestrate diverse capabilities, hardware is no longer a carrier of a single function but becomes an intelligent gateway connecting life and work, ultimately forming a truly effective closed-loop use case.
Before and after meetings, AI can do much more.
Let's return to the original strengths of the Shokz OpenFit 2 AI.
In my previous experience, the thing that impressed me most was how seamlessly it handles "recording." Wearing the earphones, I can record calls or impromptu conversations; placing them in the charging case, they can sit on a desk and capture meetings. The audio is then transcribed and summarized by software into an easily reviewable text document.
This is a mature workflow that has been repeatedly validated by many "AI earphones" over the past year. It aligns closely with the audio scenarios where earphones naturally excel, and provides a great entry point for AI to integrate into work scenarios.
But getting "one thing done" often requires more than just transcription.
When the meeting is, who needs to receive materials afterward, when the first draft is due, and when and where the next communication happens—these details, even after being processed as text, ultimately must be entered into calendars, to-do lists, and memos before they can drive follow-up work.
More critically, transcribed materials are "fixed," but life and work are always full of "fluidity."
As an editor, my schedule doesn't always go according to the initial plan. A sudden need can disrupt the original timeline; while walking to grab a meal, I might suddenly think of something that needs to be added.
If I were using the "original" OpenFit 2 AI, I'd likely have two options: either record a new voice memo about the new arrangement or thought, and reorganize it later; or immediately pull out my phone to update memos and calendars.
The significance of what LightSail brings here lies precisely in handling information orally at the first moment.
"Smart Messages" filters and delivers constantly updated communication content, "Calendar & Tasks" handles time and to-dos, and "Quick Notes" saves ideas that pop up anytime... This means that whether I'm walking, driving, or even cooking, I can give the AI commands at any time, and just by "moving my mouth" I can keep everything organized.
Speak to the earphones, and let the earphones handle it.
The logic of using voice to control workflows is closely tied to the device form of the LightSail AI full-sense wearable itself—essentially, its intelligent hub is the charging case equipped with eSIM and GPS, and the charging case has no GUI interface.
Therefore, from the very start, it prioritized processing work through voice.
Even when this platform jumped out of its own hardware and "transplanted" onto the OpenFit 2 AI, which uses a smartphone as its intelligent hub, its fundamental logic remains unchanged. For someone like me who doesn't want to constantly take out a phone, voice interaction is clearly a very useful capability upgrade.
The easiest example to understand is scheduling a meeting.
I can now simply say, "Xiaofan, help me remember: tomorrow at 3 PM, go to the next building for a coordination meeting," and the AI will proactively record the schedule and set a reminder for me—as long as the earphones are still on my ears the next day, it will issue a voice reminder before the meeting, serving as an all-day nudge.
Of course, if I've already scheduled another activity for that time slot, the AI will immediately alert me to the schedule conflict and suggest changing the time; if the meeting gets canceled, deleting the schedule via voice is just one sentence.
The same task—from creation to inquiry to modification—unfolds through conversation. This all-day, always-available calendar assistant is clearly far more intuitive than tapping on a phone.
Memos have undergone a similar change.
Everyone has probably had this experience: you think of a brilliant idea, but when you pick up your phone to record it, a message grabs your attention—by the time you finish reading the message, the idea that just surfaced has vanished.
For situations that require a "digital detox," LightSail's interaction approach is: say "Xiaofan, help me note down an idea," and voicing it generates a memo on the spot, without any interruption from other information; afterward, you can also query or modify that memo via voice. Even when I can't remember the exact wording, I can still find the record.
In my previous experience with the LightSail AI earphones, I noticed that it excels as an information funnel, filtering out useful messages from WeChat and Feishu for me. Now, the same capability has come to the Shokz earphones via the LightSail AI OS—
As messages pour in, the AI automatically filters out casual chatter and pushes only the most critical information to my ears in voice form. Different users can choose between "fast," "medium," or "slow" summarization modes based on their habits, to handle different work and social scenarios.
Whether on the way to work or packing luggage, I can still calmly keep up with the constantly updating activity in contacts and work groups.
After a series of experiences, I noticed that the core selling point of LightSail's platform-based AI approach is essentially a software experience.
If recording is responsible for capturing information, then the LightSail AI OS is responsible for calling and processing that information, building a workflow for me within its software services.
Since it is a software ecosystem, what it needs is actually simple: a microphone for inputting information (perhaps also a camera and GPS chip), a data pathway to upload information to the cloud for AI processing, and a speaker for ultimately delivering feedback.
And these essential hardware components already exist in AI earphones, smart glasses, and other smart accessories.
In other words, LightSail AI doesn't have to be tied to LightSail's own hardware. It can serve as an AI service solution and enter hardware devices from other brands.
Adopting ready-made AI, hardware gets another way forward.
There's a story in Zhuangzi: a family in Song had a special ointment that prevented chapped hands and had used it for generations for washing silk thread. Later, a guest bought the formula and used it in winter water battles, helping the King of Wu win, and was rewarded.
Technologies often have similar fates. When born, it serves a specific problem; but as it spreads beyond its original device form, it can still discover countless new uses and possibilities.
LightSail's AI service capabilities could be exactly such a "formula."
On its own full-sense wearable device, LightSail AI needs to work with the earphones, charging case, and watch to explore a relatively complete all-in-one personal AI experience. When it enters the Shokz OpenFit 2 AI, it sheds the "heavier" visual recognition and standalone internet capabilities, yet undergoes a remarkable chemical reaction with Shokz's original "record-transcribe-summarize" capabilities.
The abundant recording materials provide diverse context for schedule, memo, and message handling, and LightSail AI also achieves always-timely message announcements on these open-ear earphones that were already suitable for long-duration wear.
For existing Shokz users, they can continue to use the familiar earphones for music, calls, and recording, while also trying to delegate a few tasks that previously required taking out a phone to the new AI workflow. The hardware's inherent strengths thus gain more room to shine.
This precisely addresses the challenge many hardware brands face: the device is already well made, so what else can be offered to users?
Sound quality, comfort, and battery life are certainly worth continued refinement, but these improvements typically occur within existing use cases. AI offers the possibility of adding new use cases. Whether a pair of earphones can help manage schedules, or whether a pair of glasses can spontaneously note down what you see and hear, all determine how many moments users will think of it and use it.
After the hardware is delivered, software can still be updated, and use cases can continue to expand. For brands, this is also a key way to add more value to the device after delivery and extend the relationship between the device and the user.
In the past, a major obstacle to adding new capabilities to hardware was R&D cost—there are many tech hardware brands, but very few companies with the massive terminal scale to distribute software investments across vast numbers of devices.
For a hardware manufacturer focused on a specific category, even if the device sells well, it doesn't mean it's suitable to independently handle all the work of model orchestration, voice interaction, service integration, and long-term maintenance.
In this regard, LightSail's ready-made solution is quite attractive.
Using an external approach also provides another way to allocate R&D resources. Shokz can continue to refine its expertise in acoustics, fit, and recording experience; fitness earphone brands can also continue to improve in sensors, structure, and algorithms, while leveraging LightSail's accumulated information processing expertise to add new use cases to their devices.
This is actually a common practice in the consumer electronics industry.
Take common TWS earphones as an example: behind noise cancellation, sound effects, and connection technology, there are solution providers that serve multiple manufacturers. On top of these shared foundations, brands continue to develop their own acoustic designs, interactions, and device trade-offs. Using external technologies doesn't prevent the final devices from having their own character.
What LightSail AI offers is precisely such a model that enables professional capabilities to flow between brands.
It brings into collaboration the approach of organizing capabilities around voice interaction, as well as experience in connecting conversations, managing information, and executing operations. Hardware brands can use this as a foundation to select the capabilities that suit them and continue to define their devices.
For Shokz, the result is the combination of open-ear comfort and personal task management. In the future, if this enters other device categories, the same technology foundation may be reassembled according to different usage conditions. The more familiar a collaborating brand is with its own users, the more likely it is to discover new uses for AI capabilities.
As for what else these capabilities can do, that will ultimately be verified by the market.