腾讯新发了一个带墨水屏的 AI 硬件,设计真不错。Tencent releases new AI hardware with e-ink display, design is impressive


AI free token/AI免费积分

本文来源:https://mp.weixin.qq.com/s/a4ccu7uql6TMVJHs_fRAiw
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

总算可以公开了,心里憋了挺久的。

腾讯会议与影石联合推出了一款搭载墨水屏的 AI 录音设备,外观设计非常精致。这款小设备我上周就拿到手了,作为平时经常用录音笔的人,我也提前体验了一阵子内测版本。

腾讯新发了一个带墨水屏的 AI 硬件,设计真不错。

让我最钟意的是它内置的六色墨水屏。

大家可以看看我上面拍的图片,屏幕上显示的是我设备的 Logo。通过腾讯会议 App,你可以随时将它替换成自己喜欢的图案。

文章图片

比如上周六参加 AI Maker Summit 时,我就把屏幕换成了大会的 Logo,然后直接磁吸扣在衣领上。戴上去就像一个迷你的电子工牌。

文章图片

我的同事觉得戴着这个录音设备出去采访再合适不过了。

与嘉宾交流时,只需把墨水屏上的图案换成自家媒体的 Logo,挂在胸前即可。这样既完成了录音,又能很自然地向对方表明身份。

特别是在一些大会或活动现场,通常会有多家媒体同时在场,这时候这个设计就显得特别实用了。

另外,整个设备非常轻巧,重量仅 19.7 克,甚至比一节 5 号电池还轻。

这类设备终究需要长时间佩戴在身上。特别是夏季穿着短袖时,只要稍微有些分量,衣物被拉扯下坠的部分就会显得十分不雅。

上个星期六我全天候佩戴着它,几乎感觉不到它的存在。体积方面,它大概就同一枚小型徽章般大小。

在我看来,此类 AI 硬件最终比拼的核心无非是体积小、重量轻且省电。毕竟只有用户愿意佩戴,它才有可能不断捕捉线下的交流互动,进而转化为 AI 能够利用的 Context。

但话说回来,腾讯会议 App 此前就已上线了独立的录音笔功能。这项功能我一直都在频繁使用。

文章图片

由于苹果系统自带的录音功能依然存在不少短板。比如在录音期间如果刷到带有声音的短视频,录音进程或许就会直接中断。

这类中断往往缺乏明显的提示,往往是过了许久才察觉,导致后半段内容完全没有被记录下来。

然而,软件层面的录音笔即便做得再出色,依然存在局限性,归根结底它底层依赖的还是手机麦克风。比如在打电话的场景下,它基本上就无法发挥作用。

若是要持续录音五六个小时也颇为不便,毕竟手机的麦克风会一直处于被占用的状态。

此外,在某些特定场景下,本来就更加适合使用领夹麦。比如在进行演讲分享、授课,或是参与圆桌讨论时,领夹麦与嘴巴的距离相对恒定,收录的声音会更加平稳,不会出现忽远忽近的问题。

相较于直接举着手机录音,这种方式在观感上也显得更为专业。

此次腾讯会议与影石 Insta360 合作推出的录音硬件,宛如赋予 AI 的双耳。我个人预测,未来会有越来越多的人通过各类 AI 硬件设备,将自身真实的线下交流转化为 Context。

身处 AI 时代,我们理应有意识地培养一种习惯,去记录并保存线下的沟通资产。

周二早上我参加了一场线下活动,期间与几位老友闲聊。我分享了一点个人体悟,他们听后都深表赞同:

回想去年的 AI 机型讨论,大家口中的高频词汇还是模型。而到了今年上半年,Harness 成了热议的焦点。等到了下半年,我明显察觉到 Context 已经被推到了一个极其核心的地位。

随着模型能力日趋逼近,Harness 这套体系也在迅速走向成熟。此时大家渐渐意识到,真正决定高下的,其实是 AI 究竟能获取多少高质量的 Context。

面对同一个模型,了解你的业务状况、会议内容、文档资料以及过往经历,与对其一无所知相比,最终所能达成的效果是天差地别的。

此次补齐了录音硬件这块拼图后,腾讯会议便能够全面获取各种场景下的沟通 Context。

线上会议自不必说,这本就是腾讯会议发家的基本盘。若是在线下临时展开讨论且未携带专业设备,只需直接开启 App 内置的录音笔功能即可进行录制。

面对采访、客户拜访、线下活动这类需要持续录制数小时的场景,如今也有了专门的独立录音硬件来应对。

如此一来,笔者日常工作当中无论是正式还是非正式的沟通交流,都能接入腾讯会议,进而沉淀为 Context。

既然话都说到这份上了,我必须分享一个真实的 Case,否则光凭空谈 Context,恐怕很多同学仍然难以理解。

上周六,笔者所在团队组织了 AI Maker Summit,我从一大早到晚上始终处于高强度工作状态。

说实话前两天晚上也熬得挺晚,到了下午三点多,大脑的"内存"已经严重不够用了。不过大会现场一整天都会有大量同学前来交流,我自己也会跟不少人聊天。

这些交流中其实蕴含着格外多有价值的信息。

幸运的是,所有的沟通内容我都录了下来。大家看,总共录了将近八个小时。

文章图片

录制完成之后,我立刻就能查看这八个小时录音的图文纪要。毕竟录音硬件和腾讯会议之间肯定是打通的。

大家看下面这张截图,腾讯会议能够区分不同发言人,随后把笔者在大会中遇到的难题都总结了出来......唉,特别尴尬的是,现场团队漏打了 20 多份讲师证书,这它也给指出来了。

文章图片

或者我也可以选择其它模板:

文章图片

得益于这些录音作为基础,这两天我们团队在进行复盘时,相比以往要深入透彻得多。

活动现场一天内会发生大量事件,有些问题笔者当时能察觉,有些则完全无法意识到。如今只要具备这些沟通的 Context,我就能让 AI 进行分析。

腾讯会议本身就具备 AI 功能。比如我在元宝中要求它找出表现优异的方面。

文章图片

或者直接接入 Codex、WorkBuddy 之类的 Agent 也是可行的。毕竟腾讯会议提供了 CLI,完成授权后,便能对接到所有的 Agent 中。

下方截图展示的是我在 Codex 中输入的提示词,存在一些错别字,请大家见谅,因为我使用的是语音输入且语速较快。不过问题不大,AI 能够根据上下文及记忆自动进行修正。

文章图片

最终,它为我生成了一份 35 页的 Slides。

文章图片

组织这类活动对我来说并非首次。坦白说,以往若没有如此完整的录音 Context,等过几天再复盘时,许多现场的真实感受早已模糊。更有许多棘手问题,笔者在现场时根本未曾发觉。

不过,这些信息只要被录制下来,我就能不断地与 AI 进行推演和探讨。

近两日,我与我的团队正频繁地与用户进行会议后的电话回访。这些通话内容均被我录音留存。最终,我们会把面谈记录、电话回访以及问卷反馈等所有资料转化为 Context,以此为基础进一步与 AI 展开互动。

文章图片

遇到类似的情况,大家不妨亲自尝试一番,效果可以说是立竿见影。

我在这里做一个较为极端的设想。倘若缺少了这些 Context,那么就算我使用的是 OpenAI 推出的 GPT-100,哪怕该模型已经强悍到无可匹敌的地步,这对我而言又能产生什么实质价值呢?

毕竟它压根不了解本次大会的具体情况,不清楚用户的真实反馈,也不知晓我和团队前期开展过哪些工作。指望它在毫无背景信息的情况下帮我做复盘,就算其智力再高也是徒劳。

Context is king.

Source: https://mp.weixin.qq.com/s/a4ccu7uql6TMVJHs_fRAiw Contact me if there are copyright issues: 13951431913#139.com or leave a comment on the right.

Finally I can talk about it, I’ve been holding it in for quite a while.

Tencent Meeting and Insta360 have jointly launched an AI voice recorder with an e-ink display, featuring a very exquisite design. I got this small device last week and, as someone who frequently uses voice recorders, I’ve already had the chance to try out the beta version.

Tencent releases new AI hardware with e-ink display, design is impressive

What I like most about it is the built-in six-color e-ink screen.

You can see from the image above that the screen displays my device’s logo. Via the Tencent Meeting app, you can easily change this to any pattern you like.

文章图片

For instance, last Saturday at the AI Maker Summit, I swapped the screen to show the conference logo and clipped it magnetically to my collar. It instantly became a miniature electronic name badge.

文章图片

My colleagues think this is perfect for field interviews.

When talking to interviewees, you can change the e-ink image to your media’s logo and wear it on your chest — this both records the conversation and naturally introduces who you’re from.

Especially at large events where multiple media outlets are present, this design is incredibly practical.

Additionally, the device is very lightweight — only 19.7 g, even lighter than a standard AA battery.

Since you’ll be wearing it for long periods, especially in summer with lighter clothing, having it weigh almost nothing prevents embarrassing downward pulls.

I wore it all day last Saturday and barely noticed it. It’s about the size of a small badge.

In my view, the endgame for AI hardware is all about being small, light, and power-efficient — so people actually want to wear it. Only then can it continuously capture real-world interactions and turn them into meaningful context for AI.

Speaking of which, Tencent Meeting already offered a stand-alone voice recorder feature in its app, which I’ve used extensively.

文章图片

Apple’s native voice recorder still has limitations — for example, if you play a sound clip on your phone during recording, it might interrupt the recording without warning.

But software recorders still rely on the phone’s microphone and have limits — like not working during phone calls, or draining battery over hours of use.

Plus, for presentations, lectures, or roundtable discussions, a lapel mic is better — it stays positioned near the mouth, ensuring consistent sound quality without drift.

This new hardware partnership with Insta360 essentially gives AI “ears” — I predict many more people will start recording real-life conversations to feed AI context.

In the AI era, we should consciously preserve our offline communication assets.

Last Tuesday morning, I attended an event and chatted with friends — I shared some insights they agreed with.

In last year’s AI conversations, we talked about models. This year, it’s about harness/frameworks. Now, context is king.

As models and harnesses mature, the difference comes from how much high-quality context AI has access to — knowing your business, meetings, docs, history vs. starting from scratch.

After adding this hardware, Tencent Meeting can now capture context across all communication modes: online meetings (their core), impromptu recordings via app, and now long-duration interviews via dedicated hardware.

So, here’s a real case: last Saturday’s AI Maker Summit. I worked hard all day and night — my brain was fried by 3 PM. But I recorded everything — nearly 8 hours.

文章图片

Immediately after, I got a full transcript and analysis — Tencent Meeting identified speakers and even flagged that we forgot to print 20+ speaker certificates!

文章图片

Or I can look at other templates:

文章图片

Thanks to these recordings, our team’s post-event review was much deeper.

At events, many things happen without notice — but now we can replay and analyze with AI. For example, I asked it to find highlights:

文章图片

Or use Agents like Codex or WorkBuddy — Tencent Meeting’s CLI allows full integration.

Below is me using Codex with voice input (pardon typos — I’m fast talker), asking it to pull relevant clips from the transcript:

文章图片

文章图片

文章图片

Without this context, even GPT-500 couldn’t help — it wouldn’t know what happened at the summit.

文章图片

文章图片

文章图片

To sum up: this device isn’t just a recorder — it’s a context generator. And in the AI world, context rules. Don’t let your experiences disappear — record them, analyze them, and let AI learn from your real-life interactions.

That’s the power of this combo.

Thanks for reading — if you found this useful, share it with someone who needs better context capture in their work. Till next time!

文章图片

Integrating with Codex or WorkBuddy would also be feasible. Since Tencent Meeting has a CLI interface, I can connect to all my Agents once the authorization is complete.

The screenshot shows the prompt I entered into Codex. While there are some typos (I apologize for that; I was using speech input at a rapid pace), it doesn't matter. The AI automatically corrects based on context and memory.

文章图片

In the end, it generated a 35-page slide for me.

文章图片

This isn't the first time I've organized an event like this. Frankly, without such a complete recording context, if I were to review it a few days later, many of the actual on-site feelings would have faded. Moreover, there are many tricky problems that I didn't even notice during the event.

But once this information is recorded, I can continuously simulate and discuss it with the AI.

Over the past two days, my team and I have been frequently conducting post-meeting phone follow-ups with users. I record all these conversations. Eventually, we transform all materials including face-to-face records, phone follow-ups, and questionnaire feedback into context, which we then use to further interact with the AI.

文章图片

If you encounter a similar situation, I suggest giving it a try — the result is immediate.

I’m making a rather extreme hypothesis here. Even if these contexts were absent and I were using GPT-100, which OpenAI claims is unmatched in strength, what real value would that bring me?

It wouldn't know anything about this conference, nor would it understand the users' genuine feedback, nor the work my team and I did beforehand. Without background information, expecting it to help me with a review would be useless, no matter how intelligent it is.

Context is king.

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry