Rabbits New AI AGENT Device Just SHOCKED The Entire INDUSTRY (Rabbit R1 Device) 

This is what google home and alexa should be by now.

I don't think that the industry is especially shocked. This device seems to demand that people to a large degree replace their habit of screen touching with voice commands. That's a really big a thing to ask for. I think a lot of people don't like talking to their devices and especially in public. This thing most likely won't take off, but I wish them good luck. It's interesting when people try new things.

i can't wait to sit in a busy train shouting at my r1 through the noise of everybody else also trying to talk to their ai agents RATHER THAN JUST PUSHING A BUTTON IN SILENCE

🎯 Key Takeaways for quick navigation: 00:00 🚀 *- Rabbit AI Demo* 00:39 🌟 *- Revolutionary Tech* 01:05 💡 *- Intuitive Interface Vision* 02:02 📱 *- Smartphone Critique* 03:05 🤖 *- AI Progress* 04:01 🕵️ *- Agent Research* 05:07 🌍 *- Universal AI Ambition* 06:00 🧠 *- Learning Interface Model* 06:55 ⚡ *- Realtime Interaction* 07:01 📲 *- R1 Reveal* 08:20 ✨ *- R1 Simplicity* 09:02 ⏱️ *- Fast Response* 10:26 🎬 *- Movie Info Query* 10:55 🌐 *- Web Portal Integration* 11:30 🔒 *- Privacy Focus* 12:33 🎵 *- Music Service Integration* 13:49 🚕 *- Ride Booking* 14:20 🍕 *- Food Ordering* Made with HARPA AI

I feel Apple and Google will likely catch up and introduce their own versions in the future. It's an idea, if it works, would be very useful. But I personally am not a fan of carrying an extra device just to voice chat with it.

I”m having a hard time not already seeing this as a dated device that will just be integrated into next gen phones. I’m underwhelmed.

The R1 is equivalent to the Zune or Ipod Touch dropping right before smartphones came out.

this is pure scam bois, the pizza didnt get delivered

I'll wait for a few real-world reviews to come out; if there is one thing I've learned controlled demos paint an overtly idealistic picture of a product, however if this can do what's shown Would totally buy one and take it along. Personally would love if the device could be personalized (avatar, voice, mannerism, etc) as it would feel like a more personalized product rather than a one size fits all that most of these commercial products come off as, either way really looking forward to this.

Crazy how affordable they have made this. Really super impressive stuff and it’s exactly what I expected would happen with the devices which we will be carrying round with us in the not too distant future

So it's an analog app, or an extrapolated digital calculator. This is the equivalent of having to carry around a calendar in your pocket instead of the app being embedded into your phone as a calendar app. It's basically a smartphone w/ AI built-in that acts on behalf of all your collected previously authenticated applications.

Ordered one. I’m a big fan of Teenage Engineering’s OP-1 synthesizer, so that was a big plus for me. If I don’t get use out of it, I’m sure someone in my family will.

In terms of AI 'devices' so far, if this holds up to the demo it is by far the most impressive.

A couple of thoughts: 1. This doesn't replace apps, per se. What they've done is make the "app" hidden. It *does not* appear to be a general-purpose agent that can simply understand computer interfaces and get things done given a goal. Instead, it appears they need to "train" each action separately, which is essentially a plugin/app architecture, just behind the scenes. Looking at the video, my guess is that they are using a vision-based LLM to recognize screen elements while generating an automation script in a tool like Playwright or Puppeteer. They then give this script to the LLM to identify substitutions (variables) and then use that script as a plugin for a normal LLM. The challenge with this approach will be robustness. It doesn't appear there is any "training" happening here, but merely recording of an automation script. The LLM can probably run the script fine, as long as there aren't any strong deviations in how the app reacts vs what was recorded. I'm not seeing any "understanding" of the UI here beyond visual recognition of UI elements. 2. Early adopters are going to buy this because of the price. Few others will, because there is nothing new from a hardware perspective that mobile phones don't already provide. Expect to see these capabilities a) within apps that can directly tie into the OS like ChatGPT and eventually b) directly within the voice assistants of the mobile OSes. If they're lucky, one of the major companies buys the technology rather than outcompeting them. 3. The latency is impressive. I wonder how much of this is streaming the audio for recognition vs waiting for the user to finish speaking before generating the text. I've already thought that one improvement of voice mode in ChatGPT is that it could use continuous recognition (though that would be computationally more expensive). Also, I'd be curious whether latency remains low at scale. It's easier to do a demo where you put 100% of your CPU resources toward the demo to make it respond quickly than it is to engineer that same low latency for millions to hundreds of millions of users. 4. Also curious whether the "foundational" Large Action Model is the only model they are using. My guess is that the Large Action Model is actually a much smaller model fine-tuned at visually recognizing UI elements and interacting with them and then they're using a traditional LLM for most of the rest of the interaction. If that's the case, I wouldn't call the Large Action Model "foundational", but rather a task-specific model, which would also be significantly cheaper to train. If they truly have a foundational LLM that competes with the like of GPT-4, why release a device? As always, watching one demo doesn't give us much information and the above is all just guesswork. Looking forward to hearing reviews from people once they have the device in their hands.

I can see this being useful for us older folks who are not used to smart phones and computers. A gadget you can talk to in realtime normal language which does stuff for you will be a game changer for us. Book me an Uber. Book me a holiday. Yes please!!!! Where can I get one?

I see this totally going the way of smart watches, especially if they are "companion" devices. I would LOVE this on my wrist. Really exciting! Thanks for sharing the keynote!

its a really interesting concept, the idea of having a hardware designed for this very purpose is also very cool. However i feel the same thing can be achieved with software and a simple app as well.

Very cool! Add 6 key Braille entry and some hardware connectivity and this could really be useful for someone with vision impairments. I mean it's really useful as is. I'm excited for this.

Does it have gpt4 or other engine connected to realtime data from the internet ?

My question is how the app can distinguish various nuanced buttons in applications. For example in the Midjourney part they showed he didn't record any part for selecting different versions of the images or upscaling the images and he didn't download the image either. I don't believe it could know what those buttons do without having tought it specifically, but say if you teach it to specifically upscale image 3 will it be able to upscale image 2? Will you instead need to teach it each button and ask it step by step? Also will you have to say all the keywords in midjourney? I know you can save presets but how do you differentiate the presets by voice? I've been thinking it would be interesting to use it to video edit but that seems complicated. I could ask it to shorten a clip by a second slightly then put another clip after maybe but I'm not confident it is advanced enough. How does this even work? Do you keep your computer running and it works through a rabbit app you install? Edit: It is all done through cloud. Theoretically you could have it control your desktop remotely.

This is insane every day there is something new. I try and explain this to normal people and they just don’t understand. I’m getting one of these. This is like one of those final piece we have been waiting for a usable LLM.

Vaporware until it ships and I’m skeptical it works as they say it does. Teenage engineering gives more credit than I’d normally give but I’m still casting doubt and it’ll likely hallucinate with absolute confidence

It worked seamlessly with the choosed scenearios, we may need a generalized review and usage report to assess the productivity. I think the biggest drawback is it's only focused voice assistant, which typically not apt to the daily work activities.

Rabbit R1's performance is unmatched. The 4GB memory, 128GB storage, Mediatek SoC, and USB-C port provide ample space and connectivity options. It's compact and powerful. Loving the innovation.

It's interesting but I think Google assistant and alexa and stuff will easily meet this and surpass it soon on your normal phone. There's no reason why they can't and if anything this will add market pressure for them to do so.

this is a integration of all existing capabilities, however, it doesn’t have a barrier to lead ahead other competitors. It can only call to the extras apps if they have a open callback interfaces. The R1 device is actually an android device. This can be done just be an App alone.

So, no voice activation? We HAVE to hold the button every time in order to interact with it?

Wait...if I wanted to book a holiday which is flights, accommodation and transport, and the agent is executing on my behalf...how can this occur without breaching privacy? Booking a flight will require my personal details and passport number (as well as my travelling companions). Similar for accommodation. It will need access to payment methods. Considering the different interfaces of websites, web forms, web apps....even if I connected all of this in that portal...where is the personal information coming from? Or have I missed something obvious in the agent/action model approach?

Adding in robotic process automation (RPA) into an LLM device is genius.

It'll serve to give Google & Apple a bit of a boot up their backside, but can't see it replacing them. Might have been better to focus on an app that integrates with existing ecosystems. This feels like, if lucky, it might be the Blackberry of the mobile AI world. What might have been interesting is if they said the LAM was running locally on the device.

the thing im worried about is if theyll allow extensions made by the community if they do it would be great for regular people but it might also make devices for penetrating systems much more accessible

- [00:00] 🚀 Introduction to Rabbit's AI-powered Device - Rabbit introduces a new AI-powered device, causing surprise in the industry with its remarkable capabilities. - The device aims to revolutionize the way we interact with computers, emphasizing a natural language-centered approach. - [01:35]📱 Challenges of Existing Mobile Devices - Existing smartphones with app-based operating systems are criticized for their lack of intuitiveness. - Fumbling through multiple apps and buttons on current devices is seen as a hindrance to a seamless user experience. - [03:33] 🤖 Evolution of Large Language Models (LLMs) - Advances in large language models (LLMs) like chatbots and natural language processing have improved user-machine interactions. - LLMs have shown proficiency in understanding user intentions but struggle to execute end-to-end tasks effectively. - [06:00] 🌐 Large Action Model (LAM) and Rabbit OS - Introduction of the Large Action Model (LAM), a foundational model that understands and executes human intentions on computers. - Rabbit OS is designed for real-time interactions, hosting the LAM to bridge the gap between understanding and action. - [08:05] 📱 Rabbit R1 Device Features - Presentation of the Rabbit R1, a standalone device equipped with a 360-degree rotational camera (Rabbit Eye), touch screen, push-to-talk button, and more. - R1's functionalities include fast responses, advanced AI-enhanced video calls, translation, document handling, and an experimental teach mode. - [25:30]💲 Pricing and Comparison - Comparison of Rabbit R1's pricing with other devices in the market, highlighting its affordability at $199. - Emphasis on R1's role as a companion device, not intended to replace smartphones but introduce a new generation of AI-powered devices. - [26:39] 🤯 Excitement about Teach Mode - Enthusiastic discussion about the revolutionary "Teach Mode" that allows users to customize the device by recording interactions, providing endless possibilities. - Acknowledgment of the unique and impressive features that set Rabbit R1 apart from other AI devices.

I, for one, welcome our new rabbit overlords.

I have to say I've been a long time fan of Teenage Engineering. When I first came across the op1 i was hooked. Since then I have wanted pretty much everything they have ever created. This product looks pretty cool. Hopefully it isn't only amazing form factor design and has actual useful utility and substance. Look forward to finding out more. Edit is from a auto corrected phrase

I could only imagine how wrist-watch, car, home products or eyewear versions of this technology will look like and how soon it will be available. It will be an absolute game changer for the world providing us the most futuristic living experience. Almost as futuristic as flying cars

Very impressive, also i wouldn't want to carry an extra device with me, but instead a " Rabbit App", cause mobile phones already have the same abilities on the hardware side. Moreover i have the feeling that it won't work that flawless outside of the U.S. even if you use the device's "native" language, but with the teach mode you have at least a strong tool to adjust to your preferences - also people tend to be lazy - Everything taught to the device could be uploaded to a pool, as a kind of international knowledge base. I am waiting for the micro device implemented in my neural system in the future, haha.

It’s a nice concept but I feel that company like google and apple will create a sort of ai assistant on their technology in the near future. It’s nice to not have to worry about typing and just saying it to your personal device. It it a new way to interact with a device and it doesn’t seem too cluttered.

I'm ready to play with this, but I imagine my android will catch up pretty quick.

I see this becoming my assistant. I just dictate things to be done, and messages to be sent, and I get back to what I'm doing. Rabbit takes care of sending messages to my team, family, and friends. Seems the preorder is a round 3 currently.

Does each app need to build a specific plugin for the this device? If so, it is no better than creating an API plug-in?

After you book the uber, will you be getting your noties thru Rabbit too when it arrives? Or through the Uber app?

1:39 bro this is bs, i am not tired of my android phone. it does everything i need. its awesome. no one wants an all in one device controlled by 1 company. also dont they know that 'pocket companion' and 'rabbit' make everyone think of something entirely different

without a subscription, how will you fund the support ongoing? how will you keep up with the new LLM's in the future. what LLM platform are you using?

" I'm not sure what to do with my hands" -RickyBobby

This definitely should be a smartwatch formfactor to be always there with you while still keeping your bigscreen in your pocket.

Can this device list items on Ebay for me simply by taking a picture and then it identifies the item, fills in the item specifics, accurately pricing and correct shipping ? IF I can teach it to do this then it would be an instant buy...

It is truly a great device; booking an Uber by simply giving directions is absolutely incredible. For home use, this device is excellent. However, it may not be able to replace Android or iPhone entirely because relying solely on voice commands might be a bit tedious. For public place usage, people are likely to prefer a graphical user interface (GUI). Nevertheless, a combination of R1 and Androids/iPhones could be a great option

What is the max token count? If you want it to do a multi step task, how well is it going to perform. Let’s say I want it to “buy a stock in my brokerage account and pass a buy signal to my discord group” when the stock is bought. I’m curious what it can do with these more complex tasks.

I think all the functions in this device can be realized through the mobile phone.

This is a start of a new era for technology using ai and I’m loving it. Maybe someday making games and social/business apps with front and backend coding in minutes can be achieved through ai.

I imagine a near future where everyone is talking to their device and Im just annoyed af

You can already play any song by using voice control. To order food, you usually want to SEE what you're ordering, and for the trip to London, you would want to LOOK at the place you're staying at. Idk what problem are they trying to solve here

ABSOLUTELY SOLD! I can imagine if they made a future version that integrated basic phone services like calling and text, and a more fluid and appealing AI virtual avatar and voices, i would literally throw my phone away and just use this. i can use all the apps on my tablets or computer, but to have true AI that can complete actions instead of showing me lists or just taking me to another website is amazing

There's no dodging that this thing has got maybe a year of operational usefulness. A smart phone with hardware based around this idea will find a huge audience, and it's a fair bet most or all existing phone producers are already working on it. Maybe if it comes paired to a smart watch or something it could find a longer lasting niche.

Very impressive how it interacts with my Google glasses in my flying automobile

Any chance of a standalone RabbitAi App that can be downloaded? In case people don't want to carry this and a phone at the same time.

Their method of achieving actions is actually really genius It looks like they're basically just doing an action while describing it and then having an LLM feed it back to their action model. That teach mode is seriously underrated and is basically its number one feature. You could use this to automate everything from booking a flight to penetration testing and security in Kali Linux or setting up complex network configurations in a firewall. This kind of action model is the next step on the path to AGI. I see people in here saying that phones can accomplish this in other companies will do this but what I think they're missing is the complete package that is being offered here. First of all, $200 is a steal. More importantly though, this is built to replace phones and while it can only do what phones can do now the foundation is groundbreaking. They are imagining our world in which we don't even need physical user interfaces because things are designed to integrate deeply with AI but the foundation is built to work with current systems as well. The specifically mentioned it should be able to work with any user interface but it seems like the goal is to create a whole new way about thinking of UI. Even little details like the fact that it's a physical button both to interact with the AI and to confirm choices shows that they're putting a lot of thought into this.

Fifteen years ago, your GPS was a device you mounted in your car. Garmin. Tom Tom and so on. They looked like a phone does today. Now, nobody would use such a clunky thing. This is the same. In a few years from now, this will be a big assed brick that nobody wants. But its cool for the interim.

When doing things like ordering food or ride share, I assume it's doing those things over wifi? Which means through my existing cell phone data service? Which means I'd experience the same limitations as I do with my cell phone?

I feel this kind of device will evolve to help a lot of careers that need some sort of analysis, for example data analysis, corrections, even radiologists or pathologists in medicine they have a chance to enhance their career by this kind of assistance.

Batman’s “supercomputer”is finally becoming real 🎉 The concept of having a layer above the UX to mitigate the frustration that comes with modern devices is genius. Also love the inclusion of analog controls. already a LAM fangirl ❤

Holy cow.. I haven't seen a device I've wanted this much, in nearly a decade. And only $200? :O I can't afford to preorder without seeing hands on reviews, but I'll certainly be considering it in the future!

Everyone is saying that it could just be an app but the rotational camera is what makes it special even though when he point his camera on to the guy they just cut away it must've been malfunctioning at that point lol

I don't understand why it has a button that you have to hold down. I'd like to use this if it was sitting on my desk without having to pick it up.

This is obviously cool ... but could be done with "an app from an app store" without a new device. 😂😂

I have only one question. (Well, I've got many more 😅) How much of these features can be used when it's entirely offline? Of course, things like booking a hotel or fetching current stock prices can't be done offline. But what about other features shown in this demo?

Rabbit R1 not only excels in communication with its mediatek chipset but also stands out for its compact size, fitting comfortably in your pocket for on-the-go adventures

I think instead of making a new device they should have to make a seperate app for it because all the camera and scree ,mic are already have with our phone But probably for older ones its is one of the best product to consider

Jesse Lyu needs to be adopted by an Italian family to learn how to speak with his hands.

Everyone saying apple and Google can do this, well, they haven't, this device Basically eliminates the need for apps, thus the need for phones. I am sure they could have put a super high end screen and camera, but they decided to keep it affordable, Wich is a very smart move, they can always improve the hardware and make expensive models. I for one I am pretty sick and tired of looking at a screen all day to get this done, with this device I can get back to experiencing real life and getting things done mostly with voice commands.

17:40 ahahah, he rick rolled us 😂😂

Reminds me of the two-way pager. Worked great, but once phones baked it into their devices, it was over. Same thing here.

Nice concept but limited to few basic tasks as an everyday phone. The current phones have a huge array of functionalities that's build deep in to the ecosystem. This can be compared to the minimalistic Nothing phone. Pretty sure Apple and Google are already working on similar features. We will start seeing them when its ready for consumer level adoption.

This should be in size of smartwach, that would be much easier to wear. Right now its another device to put into my pocket.. Im waiting for superinteligent AI smartwatch

Now, if this develops to the point where you can buy a singular device yet just replace the hardware with affordable components when new things update and change so it can stay relevant for large amounts of time(with open source visualization for continuous transparency and outright ultimate security), THATS what I would drop money into. Also proper voice fluctuation.

"Rabbit, please give me back these 30 mins"

I’ll wait to see what Apple announces for updates to Siri in June. Cool to start seeing these devices though, I thought about buying one when the said the price, but who knows what else there will be by April.

I also feel this type of functionality can easily be incorporated *into* an iPhone. Why not?

This demo made my vaporware sense tingle. Let's wait and see how it's performing in easter.

We are one step closer to Star Trek, “Computer, take me to the bridge.” XD

"If you look at the top ranking apps on app stores today you'll find that most of them focus on entertainment. Our smartphone has become the best device to kill time, instead of saving time." ~Jesse Lyu, CEO Rabbit

I'd just like to go on record saying I welcome our new AI overlords.

cool camera design, would like to note that its a missed oppertunity to open or add a glass on top so the camera has additional angles

Can this thing make me an excel sheet based on data that i showed through the camera and also point if there are some error in the data?

Next step to AI use for public. But, Smart Phones can go in this direction themselves.

How is this different from any internet connected LLM?

Imagine creating scrolling "button" and still use screen to scroll :)

I think Google or Apple can work to make a single app to do all these tasks ( may teach their assistant app) , and they integrate all those tasks into single interface, while we use the same mobile device. I never thought of second device as of now besides our mobile phones.

Some things to think about: - the UI is a bit stupid: that rabbit does nothing for me. To make it useful, it needs different states/emotions and an interactive capability based on context of user and/or surrounding and tasks it is performing. A huge challenge in and of itself. If this isn't feasible, it needs to go. - Speech seems quite an inconvenient mode of interfacing/inputting in a bazillion public situations. I'd rather type and remain quiet. I know they have a keyboard. We'll see how this works with it. - This is only useful for me if the hardware is really good: camera needs to be excellent and the device itself robust and water proof, ideally. - training it for tasks only works for me, if after one single training attempt it works flawlessly. Otherwise, I can't be bothered to be their paying customer and LAM-trainer at the same time.

join the robot religion today! ❤️

I love it I think it’s neat. It definitely won’t be for everyone. I think if Jesse made it into a brand fad like releasing different colors, voices, and other brand marketing strategies it would be a hit. I love the price too

I have an interesting question. Could you use teach mode and show it to just research any topic via RU-vid videos learning sites ect to just teach itself?

"it's funny you know all these AI 'weights'. they're just basically numbers in a comma separated value file and that's our digital God, a CSV file." ~Elon Musk. 12/2023

It’s the $199 price that got me to take it seriously tbh. The AI Pin from the ridiculously hyped “Humane” is the worst value for price. But tbh, I don’t think a phone obviously can be discarded. Pretty good.

Love that it's a separate device, lowers hacking attack surface. Great to just take up, tap a button and get an uber. Way better than an app on the phone!

They need to teach it mixing and mastering in Logic Pro and pro-tools so it can finalize and make ready for release and master the composition in all formats at broadcast levels. That should happen asap. There is a ton of very talented musicians and artists that can’t produce anything at levels the industry demands. It’s time that changes forever. This could also greatly simplify editing process for film . I’m certain apple and intel know this is going to eat them alive if it doesn’t happen by the next Apple event. Cause I want one of these like yesterday. 👽✌️

I think their teach mode is REALLY cool and they would probably make a lot of money if they sold that engineering to one of the larger companies so they can integrate it into their system. Having a second device (phone + rabbit) to keep track of is probably not what people are looking to do. The ability to talk to a device and call up any task is VERY useful, but having to go through the steps of teaching a device a new task will limit who will find rabbit useful. I am thinking of the much older generation and the technophobes who are intimidated by electronics. If they can refine their LAM and make it completely intuitive requiring no manual "go to your computer and do a teach session", then they quite possibly will have explosive sales. The older pop. and pop. of technophobes is quite large. Many everyday people are behind in what they know and understand about AI. Apply the K.I.S.S. (Keep It Simple Stupid) principle for the masses and you have a win because it will appeal to everyone.

How often do you book a vacation for the entire family vs responding to an email or text message - yet that’s what they chose to demonstrate (to the exclusion of the far more common messaging). The device also seems to require that you have the ability to verbalize your request accurately. I would have liked to see them edit a misspoken word and/or use the keyboard to modify a request. They should have also talked about battery life, changing the voice, data plan subscriptions, taking a picture or video, screen resolution, etc. I didn’t hear them mention a GPS or any navigation capabilities. I think this will be a niche product that I may pick up out of curiosity when it drops below $99.

Love your video!

Very impressive. This device made April 1 arrive early.