動手玩：亞馬遜最新AI穿戴裝置Bee

Topics

Latest

AI

Amazon

Apps

Biotech & Health

Climate

Cloud Computing

Commerce

Crypto

Enterprise

EVs

Fintech

Fundraising

Gadgets

Gaming

Google

Government & Policy

Hardware

Instagram

Layoffs

Media & Entertainment

Meta

Microsoft

Privacy

Robotics

Security

Social

Space

Startups

TikTok

Transportation

Venture

More from TechCrunch

Staff

Events

Startup Battlefield

StrictlyVC

Newsletters

Podcasts

Videos

Partner Content

TechCrunch Brand Studio

Crunchboard

Contact Us

Hands-on with Bee, Amazon’s latest AI wearable

In early tests with a review unit of Bee, we found the device itself was easy to use. It’s just a press of a button to turn recording on or off. In the app, you can configure whether a double press bookmarks a section of the conversation, processes the current conversation, or both, and you can set whether a press and hold gesture lets you leave a voice note or chat with the AI assistant. (Bee’s companion app currently reminds you to enable voice notes, so we did.)

Like many other AI products and services, such as Plaud, Granola, Fathom, Fireflies, Otter, and more, Bee can listen, record, and transcribe audio conversations. Where it differs is that instead of offering an overview or a raw transcript, it segments the audio into sections and summarizes each part. For instance, an interview might be segmented into sections like the introduction, the nitty-gritty product details, an overview of industry trends, and whatever else you may have talked about.

Each section is tinted with a different background color for easier differentiation as you scroll. You can tap into an individual section to see the exact transcription.

It wasn’t immediately obvious how to label the speakers in the app — we learned we could tap on a segment of the conversation to confirm if we were the speaker, but this fell short of other professional AI transcribers, where each speaker could be labeled. In addition, Bee discards the audio after transcription, making it a non-starter for use cases where you need to play back the audio to ensure accuracy.

That said, Bee isn’t necessarily meant to be a work tool. Amazon sees this as an AI that can live alongside you as you go about your day. By integrating with Google’s services, Bee can tie a recorded conversation to a task. For instance, after meeting someone at a conference, it could suggest that you friend them on LinkedIn or research their product.

You can also leave yourself voice notes, as an alternative to writing something down in your notes app, for instance.

Another section in Bee’s app lets you look back at past days’ memories, while a “Grow” section will offer insights the more it learns about you. You can also confirm and add to a “facts” section about yourself, which is somewhat equivalent to other AI chatbots’ ability to remember things you discussed.

Amazon says it will be shipping more features for Bee in the year ahead.

Bee isn’t always listening by default, which is why rival wearables like the Friend AI pendant saw backlash. Instead, you’re meant to ask if you can record someone’s conversation (unless at a public event of some sort, where recording is already expected).

When you do record, a green light turns on, alerting others to the fact that the device is in use.

Bee’s sports band was a little flimsy. The band fell off twice while being worn, both times while just sitting and not moving the hands much (like in a taxi). We have not yet tested the clip-on pin, but it feels more sturdy.

Overall, the mobile app’s design is far ahead of the apps Amazon has built in-house, like the Alexa mobile experience, and it’s easy to use. But the premise that we need an AI specifically to record conversations to learn more about us is still largely untested. Is there a world where such devices make sense for consumers who aren’t recording in professional settings, like meetings and interviews?

Plus, if AI listening devices go mainstream, there will also have to be some sort of cultural shift in terms of what’s appropriate and what’s not. Today, it’s somewhat looked down upon to record video of everyday people going about their lives, even though it’s technically legal when they’re in public; similarly, it may be considered tasteless or gauche to record audio with an AI device if you don’t first ask permission.

Not everyone will abide by that social contract, of course, which could see people self-censoring their speech in public.

At CES, for instance, we were chatting with a rep at the Soundcore booth. When they liked something I said about a competitor’s product, they joked, ‘Say that louder into my microphone,’ pointing to the already-recording AI device subtly pinned to their shirt. It was an odd experience to realize that everything said in the real world could one day be “on the record,” whether you consented or not.

Bee’s traction — or lack thereof — will help Amazon determine if that’s a world that consumers actually want.