Rabbit R1: The First Personal AI AGENT Device NO ONE Saw Coming (Look Out, Apple)

Let's watch the new Rabbit R1 demo and discuss!

UPDATE: I had to remove the sound at 17:46 for about 20 seconds because it was a copyrighted song.

Enjoy :)

Join My Newsletter for Regular AI Updates 👇🏼

My Links 🔗

Media/Sponsorship Inquiries 📈

Links:

[matthew_berman]

I purchased one! Do you want to see a full review when I get it?

[..]

When I first learned about this, I thought it was very cool, because I thought the device was built to run AI on it, locally. Then I learned what the device ACTUALLY is. A phone with one app and a (inherently dubious) Cloud(tm).

[tensiondriven]

Note this /is/ an Android device, it runs ASOP. A couple points that were glossed over in the presentation and most videos I've seen covering this:
- It's not a fully local LLM, so you're using their cloud, and that cloud is going to be expensive. All the promises of privacy-first will disappear when they need to show profit.
- ChatGPT isn't slow because of cloud. Look at how Home Assistant does voice, or how the ChatGPT iOS app does voice; in both cases, it's pretty speedy.
- How does the rabbit interact with a PC or android device? Is it done in the cloud on emulators? That's not private! This point seems to have been glossed over too.
- also, just had to add, Teenage Engineering did the industrial design. I ordered one half-way just for this reason :) I hope it can be rooted so we can still get some mileage out of it in two years when it gets supplanted by Siri 2 or whatever else comes next.

[TrevorLaVigne]

It's like a magic 8 ball, but with added government surveillance

[laserboneful]

I remember a different personal device called a rabbit. Hopefully consumers don’t get confused which is which

[havoc-ado]

The presentation feels like the intro cutscene of a game that's based on a robot apocalypse

[MrOwenEaster]

The irony is that this should have just been an app.

[CoyotePark]

Legends say he's still holding his hands up

[jeffreyjohnson]

As a pilot the LAM concept would be insane for finding the best flight itinerary with the best chance of getting on for non-revving or jumpseating and considering connection times, distance between gates, etc. Just being able to pull the rabbit out and tell it what I need would be game changing

[ekkamailax]

For simple actions like “play a song” this works fine. But for more complex stuff like booking a trip or an Uber I think people want to see an interactive user interface and have total control. Just imagine this AI making mistakes and now you’re wasting time arguing with it when you could just open the Uber app and select the exact options you want.

All that’s really needed is for Siri’s speech recognition to get better, so you can say “hey siri open Uber” and it immediately opens the Uber app for you.

[CharveL]

I appreciate you early adopters beta testing this. I'll hang in there for a few months after release so I can decide after the copium runs out on the showstopper problem with this becomes obvious. Learned my lesson with a couple other similar "ground breaking" devices.

It's always the thing you don't see coming, like say requiring a subscription after this startup burns through it's capitol maintaining cloud service support; or it gets spotty service because the antennae was cheap; or it stops working if exposed to high humidity in your part of the world and the company stops responding to the flood of service requests; or the camera only recognizes the veggies in your fridge if they aren't staged like in the video...that kind of thing.

I do want to...want...it, but I don't mind being a couple months down the adoption curve.

[tortellofer]

rabbit r1 could be an App on an iPhone as well. There is no need for another device. In addition, Siri and other language assistants will develop all that capability natively. Eye-glass computing with this functionality would be even more advantageous than another device in your pocket.

[PMX]

"No subscription" but their LLM runs in the cloud (expensive) and the interaction with apps is done in emulators running, again, on the cloud (expensive x 2). So, either they'll start charging for a subscription in a few months or they'll go bankrupt and the device will become a (very nice looking) paperweight.

[theblockchainsolution]

Lol on the copyrighted lyrics I wondered what happened to the sound I didn't worry too much about it because of the segment that was happening. Thanks for posting this. It was very interesting.

[davideide]

Wonderful idea this is going to help everyone save so much time thank you

[rjnash]

Depending on the amount and the speed of sales of the device will also depict how long it'll take for them to practically throw in the towel and release the app for Android/IOS (especially as in a few months MAX, we'll all see a flood of such AI apps all over the place! :)
100% if they have any sense, there's already gonna be a strong beta version already ready to roll out right now though (that they're gonna keep primed and ready as a fall back).

[bladestarX]

The concept is promising, but it appears to be transitional or temporary, and I am skeptical about its potential to become mainstream. The barrier to entry for mobile companies is excessively low. I cannot identify any functionality this device offers that cannot be executed more effectively with a modern phone. This device encounters the same challenges as mobile gaming devices and tablets. Most individuals prefer carrying or using a single device, with their phones serving as the ideal choice. I believe it will perform well until Apple or Samsung incorporates it as a default feature or capability in their operating systems, or until a dedicated app is released. This situation brings to mind the first person who introduced an app enabling the use of a phone's camera flash as a flashlight. Why carry a separate flashlight when you already have a phone? I anticipate that in the future, every device, such as TVs, cars, phones, and individual applications, will come equipped with built-in natural language models. They will possess integration capabilities and be capable of executing these tasks seamlessly. These models will be small, local and limited to their purpose. Having a be all and know all device or model is not ideal.

[davestorm]

We need to be focusing on personal AI systems (distributed) vs centralized AI systems. To prevent an AI overtake, we need to look at personal AI systems that are loyal to the user and even protect them from the potential of malevolent/destructive AIs, like having a guard dog at your side 24/7. Tasks would include everything Rabbit does to background processes protecting privacy and pro-active security (including verifying identity).
A personal (and loyal) AI system would also recognize and prevent bad government actors (and other large entities bent on control), be they human or AGI, from doing bad things on a massive scale.

[flyingsalmon]

Jesse is spot on regarding the limitations of today's "smart" devices...that just listens and retrieves information or creates images but doesn't actually DO much else for you. r1 is still a patchy solution from intent->action however as there's so much incoherence/constant changes/inconsistencies with interfaces across apps. I see the idea of intent-based operating systems being a HUGH disruptor in the operating system ecosystem when coupled with action-based AI models. Today's OSs are just cumbersome for most people requiring too many inputs and jargonized commands. Your commentary and summary was great at the end (as always)...appreciate it!

[ApollosChariot]

I can see this product being great for kids! It's minimal, simple, and less distracting than a traditional phone loaded with games and apps. And would help them stay away from the dangers of the internet and social media unless it was allowed by the parents.