Quick feedback if the founders are watching this: Please allow controls on your keynote video. The video production quality looks great. But I am not watching it if I don't know if it's a 30s one or a 30minute one.
Edit: My feedback is that it's too slow. I wish the features were shown off before I had to get a layman's education for several minutes. I'm also really concerned that his uber will arrive before the pizza.
I like the device. It seems easy enough for my elderly father to use, and he won't lose it as "one of the websites" he can ask for help. There's something important about the focus a physical device brings; he can remember how to use it, where it's at, etc. The GPT UI I built him gets lost in his bookmarks, and he forgets it exists.
A while ago I came across a website with a video player that gaslights you with a logarithmic progress bar. It had me fooled for a while into thinking the video has only 2-3 minutes left, before I finally clicked the pop-out button in Firefox and realized I was only 20% of the way through a 45 minute video.
Is there a Firefox extension for android that can take over the js crap that people call video players and give something decent? I know there are open in x but that isn't seamless. I just want a video player that is standard across the Board.
Like safari on iOS used to turn any video on browser to a standard video player window with normal controls.
Not the same website, but the video player was made by a Brazillian company called VTurb and they're using the same logarithmic progress bar on their front page. If you inspect element it's literally called "smartplayer-fake-bar"
Also please describe it in pictures and words too. Video only really limits where I can watch this. I’m in a waiting currently and I’m not about to just play a video like some sort of psycho.
> I’m in a waiting currently and I’m not about to just play a video like some sort of psycho.
Funny enough, I think them releasing only a video perfectly aligns with their product and why I think this will fail. They seem to have completely overestimated how often people want to read or write and underestimated the number of scenarios where needing to talk or listen to my phone is inappropriate.
I'm assuming if you had this phone, you'd feel like just as much of a psycho if you had to dictate every command to it while sitting in that same waiting room.
Yeah, the price difference between this and Teenage Engineering's own products is wild! It's worth noting the Playdate (by Panic) is designed by Teenage Engineering and also just $200. https://shop.play.date
I need to start my teenage engineering collab collection. Such wonderful and refreshing experiences. I’ve used my friends playdate but never took the plunge and it’s wonderful, haptics and everything
That seems unlikely given that both Apple and Google have been employing the world's top ML scientists for years, have unlimited budgets, have better access to customers than any startup (to find out what customers want), and ... need I go on? Yes, it's nice that "Rabbit" is exploring this area and being innovative, but unless their particular take on mobile phones catches the world unexpectedly by storm, nobody will mourn their passing when the money runs out later this year.
1. OpenAI is building in a brand new space. Mobile phones are well established. LLMs as a product for consumers are brand new. While it’s true Rabbit are trying to merge LLMs with mobile, the elephant in the room is the mobile incumbency. The existing players just have to add LLMs to their existing dominant platforms and Rabbit is done for. OpenAI, on the other hand, was unopposed launching ChatGPT. They had a genuine technical edge and consumers were hungry for it. Is everyone hungry for Rabbit’s concept? Give me a break.
2. OpenAI raised an order of magnitude more capital. Rabbit’s $30M isn’t going to get them much farther than a prototype device. My impression is the founder here managed to convince some VCs to give him money during the boom times and leveraged the generative AI hype train more recently. But where is he getting his next round? The one that he will need to actually make phones at scale. That will cost billions ultimately, and the incumbents own the supply chain he needs to access. His effort is all but doomed.
3. OpenAI’s formula was easier for a startup to master. All they needed was money for the best AI engineers and scientists and money for GPUs, and they could create a blockbuster product. Rabbit needs the top engineers as well as extensive capital for manufacturing and distribution. There is a reason that hardware favors massive scale and a reason why hardware startups tend to focus on pinpoint innovations. The energy barrier is extreme.
These are three reasons why Rabbit is in an entirely different situation than OpenAI was circa 2022.
> That seems unlikely given that both Apple and Google have been employing the world's top ML scientists for years, have unlimited budgets, have better access to customers than any startup.
Both of those companies had it handed to them, like, literally got completely smoked by OpenAI, a company with a thousand employees +/- in San Francisco. The giants are incredibly vulnerable, just like the giants that Google and Apple disrupted such as IBM, Yahoo, AOL, etc.
It's a lot harder for startups to win over large companies in hardware, even with a superior product, since shipping hardware is tremendously capital-intensive. An example that comes to mind is Pebble, which had an excellent smartwatch that worked way better than the one from Fitbit, but nevertheless ran out of money and got sold to the latter.
I think an acquisition does seem pretty likely actually. Google still has yet to make assistant do anything interesting, despite leading the way on the research side of things. Also keep in mind that they lost a ton of AI talent to startups.
> That seems unlikely given that both Apple and Google have been employing the world's top ML scientists for years, have unlimited budgets, have better access to customers than any startup (to find out what customers want), and ... need I go on? Yes, it's nice that "Rabbit" is exploring this area and being innovative, but unless their particular take on mobile phones catches the world unexpectedly by storm, nobody will mourn their passing when the money runs out later this year.
I worked with one of the (many) teams at Microsoft who worked on Cortana.
The way the team leader explained it to me is that Cortana could do a lot more, but internal corporate politics prevented it. Rather than implementing the best solutions to user's problems, they had to do things like ensure Bing search handled certain results, to make sure that team stayed happy.
Or to take it to the extreme, if someone at Google came up with a device that directly beamed 100% correct search results into your brain, Google would never release the product because of the loss of search ad revenue.
I don't see it. Base model Pixel 7 can be found in multiple places for $250 right now, so you're not coming in much cheaper than that, but more importantly, I don't really see a whole lot of demand for a second phone even as a cool toy. This feels like the type of thing people would buy, play with for a week or two, and then it would end up in a drawer never to be used again.
At the moment your right, but as Google starts to implement LLM assistants into their product it's going to eat this things lunch. I would guess maybe 18-24 months max before Google assistant can do everything this thing can and they already have a phone at a similar price. Once that happens, I just don't see what niche this fills into other than "it's not Apple or Google".
My thoughts exactly. I'm willing to give it a try because it's exactly how I imagined a companion device for intelligent note taking. Given that it's a collaboration with TE, I trust the build quality. My only concerns rest in the lack of a subscription given the cloud reliance on the OS, and as I'm sure many others feel, security with a fledgling device and it's interactions as an agent of you is definitely an important consideration.
"Doomed to fail," probably. I certainly never expected to see a product so unabashedly quixotic this side of a zero-interest-rate era. But man, what a crazy swing for the fences this is.
By my reckoning, this approach is probably ten years ahead of its time. I simply cannot imagine that the tech is there right now to make this nearly as seamless as it will need to be to actually supplant the UI paradigms we have today, and it's going to take billions invested by the duopoly to get to that point.
But I really do think this is at least a fuzzy picture of where we're headed. Your iPhone in 2034 won't look like this, but you'll likely be able to trace some things back to it. There are a lot of pieces missing that we barely even know are missing yet, but it's incredibly exciting to see a startup try to jumpstart a step change like this.
Idk, I think an iPhone in ten years will look a lot like it does now just with better Siri. Better Siri will open up a lot of things and it’ll be wonderful but we’ll still want a screen (we’ll still interact with the world largely via reading) and an iPhone just looks like a screen.
Apple and Google presumably know about LLMs so I don’t really see this device being very influential.
Maybe. That'd be interesting I suppose. Those Meta smart glasses do look more usable than anything on the market previously. Maybe because they omit screens.
Perhaps at some point technology will allow glasses to have screens that are functional without making you look like, as they called them ten years ago, a Glasshole. Ten years may be enough time.
Mobile ads and app stores make Apple and Google too much money for them to truly revolutionize mobile phone user interfaces.
Anything that even threatens those business models will be strangled before it gets released.
IMHO Microsoft has the most to gain here. If they manage to turn Apple and Android into platforms that interface with MS made/ran/hosted LLMs, MS wins the day.
The problem is monetization, these LLM services are expensive, and telling users they need to pay another $30 a month to use their phone will be a hard sell. Hopefully LLM hosting costs go down, but if hosting LLMs becomes too affordable, MS's giant war chest of $ and hardware become less of an advantage.
The other danger is Apple/Google throwing a bunch of restrictions in place, but all the anti-trust scrutiny they are under right now could give them pause on attempting that.
to me the pitch is compelling. i don't want to use any more fucking apps -- i want to talk to my computer, like i talk to chatgpt. there's real opportunity for a revolution here.
i don't think apple or google are well-positioned to build this revolution, because they are too conservative and too bought-in on the old interaction model.
(if jobs were still around, different story. alas.)
Google's primary revenue driver is search ads. I'm sure IAP fees and web advertising don't hurt either. If you don't do searches, browse the web or use apps, where does the revenue come from?
Even if Google builds something like this, gets market share and sells at a profit, it might still be a net loss for them because of all the advertising money and click tracking data they won't be getting.
Google definitely has the expertise to build this, but they also have an extremely risk-averse attitude resulting in layers upon layers of bureaucracy, and this product is literally cannibalizing their most important markets. I can't see this going over very well at all the internal reviews.
Apple was a lot leaner than Google when they build the iPhone (which was also cannibalizing one of their main products, the iPod), but they still had to make a completely independent team with no oversight except for Steve Jobs to get it right.
We also know how bad Google is at hardware. They seem to have gotten marginally better in recent years, but Pixels are still far from successful, even in countries where they are sold, which honestly isn't that many in the first place.
I'm very bearish on Google in this fight. Apple probably stands a much better chance, they've already proven they can cannibalize their own products, they need to do far less of that in the first place, as most of their revenue comes from hardware sales and subscriptions (which is perfect for something like this), they do have a lot of hardware expertise and they've already made major steps towards AR glasses, which might end up as a better form factor for such a device than a phone. Their pro-privacy attitude might be an impediment, this stuff works a lot better if you run it on a beefy GPU in the cloud instead of a tiny, battery-conserving chip in your phone, but I hope they find a good enough compromise.
The fact the Vision Pro is passthrough VR instead of an AR screen on glass (as in when the battery is dead you see black, not see the room with no AR) says that it's far away.
No one, including us, will be able to use personally identifiable information (your name, phone number, email address, etc.) for any purpose other than serving you.
Is there a specific paper or something you can point me to? Or are you talking about like llama.cpp? Because I thought that referred to the fact that it was originally one c++ file named llama.cpp?
Smart phones are closer to smart cameras. I use the phone app maybe once a month. I'd argue it might be viable to build a smart phone that has no phone.
Small enough that I would carry it in my shirt pocket or back pocket for anything. Looks very slick and at a very good price. Will buy three for my siblings this christmas.
I read the comments before I watched the video and I'm actually way more optimistic about this now than I was expecting to be.
Yes, google or apple will probably ship LLMs on their phones eventually and make an overall more compelling product than Rabbit _if_ you're only willing to have one device.
Considering the relatively low price (and assuming they actually deliver it in a timely fashion) I can see this being useful as a pocket assistant between now and whenever google et al eventually catch up, and potentially past then if rabbit keeps innovating.
I see this as less of a competitor to a smartphone, and more of a competitor to a Light Phone [1]. Seems like it might help folks avoid doom-scrolling in a similar way, while providing a higher level of functionality.
- the keynote video isn't scrollable behind the giant "share" button
- the text on the keynote page is incredibly difficult to read (small text + very low contrast)
- the homepage is full of dizzying animations that don't tell me anything about what the "phone" is actually supposed to do
- "How it works" is just a set of animated cards. Once you figure out they're clickable (hoverable?), they spin around to reveal... half-formed ad copy? I wanted to know how it worked.
I'm willing to believe this is a real product only because of the "research" page, but it feels more like an aspiring web designer's portfolio piece. The flashiness of it all makes it really difficult to figure out what anything actually is.
I only skimmed the video, but I have to agree. The biggest problem I see is that I don't want to talk to my phone for a lot of things. If I'm having a text conversation or trying to look up something on the internet while I'm in a crowded room, I don't want to use my voice for that. If fact, while there are some things that I would find this helpful for, I would say the majority of the interactions with my phone are things I would actually prefer to have a keyboard for.
While this thing does have a touchscreen keyboard it seems to me like the biggest selling point of the phone is actually something I don't want to use most of the time. I can't really see this succeeding for that reason.
These demos are always the same: “<Product> find me the closest pizza”
Who actually interacts with a device like that? Is it an important problem enough to be paying $199 while I can do the same thing (often even better) with my phone?
No. I could go deep into "rare language, weird slang, local dialect" hole, but simple "No" would be enough. Their website is exceedingly non user-friendly, too
I think what’s impressive here is their approach to “Teach Mode”, which essentially teaches the average consumer how machine learning works. I think that’s already a much better approach than Humane Ai and provides a better path long term for where this could go. This approach of a Large Action Model that can control any of your apps by watching you use them is a unique implementation for a hardware device. The hardware isn’t super impressive but the approach has staying power. It’s always way less invasive than Humane AI pin, it’s deliberate usage.
I'd totally buy a ChatGPT+ device for my 7yo and I imagine people around the house would use it quite a bit.
Not sure I want to replace any other actions yet. Probably someday, but touch + apps is actually just incredibly good. A sub-1yo can start learning to use a smartphone / ipad, and a 6yo can put together an uber eats order. I'm not saying it can't be improved on, and I fully want a competent voice assistant in my phone. But it's gonna be real hard to do better than what we already have in our pockets on the things they're already good at.
So this presentation fell flat is for me are when it was talking about doing things pretty much everyone does just fine on their phone. But it was very exciting to imagine a ChatGPT+ device sitting around the house for anyone to pick up and interact with.
546 comments
[ 3.1 ms ] story [ 338 ms ] threadIt's a vimeo embed
Edit: My feedback is that it's too slow. I wish the features were shown off before I had to get a layman's education for several minutes. I'm also really concerned that his uber will arrive before the pizza.
I like the device. It seems easy enough for my elderly father to use, and he won't lose it as "one of the websites" he can ask for help. There's something important about the focus a physical device brings; he can remember how to use it, where it's at, etc. The GPT UI I built him gets lost in his bookmarks, and he forgets it exists.
Like safari on iOS used to turn any video on browser to a standard video player window with normal controls.
Funny enough, I think them releasing only a video perfectly aligns with their product and why I think this will fail. They seem to have completely overestimated how often people want to read or write and underestimated the number of scenarios where needing to talk or listen to my phone is inappropriate.
I'm assuming if you had this phone, you'd feel like just as much of a psycho if you had to dictate every command to it while sitting in that same waiting room.
This isn't a phone though right? The title seems wrong.
You've got absolutely insane market saturation by two major players that have some of the most advanced AI teams on the planet.
Almost nobody is going to buy an "LLM-first phone," they will just wait for Apple or Google to put an LLM into their existing phone.
1. OpenAI is building in a brand new space. Mobile phones are well established. LLMs as a product for consumers are brand new. While it’s true Rabbit are trying to merge LLMs with mobile, the elephant in the room is the mobile incumbency. The existing players just have to add LLMs to their existing dominant platforms and Rabbit is done for. OpenAI, on the other hand, was unopposed launching ChatGPT. They had a genuine technical edge and consumers were hungry for it. Is everyone hungry for Rabbit’s concept? Give me a break.
2. OpenAI raised an order of magnitude more capital. Rabbit’s $30M isn’t going to get them much farther than a prototype device. My impression is the founder here managed to convince some VCs to give him money during the boom times and leveraged the generative AI hype train more recently. But where is he getting his next round? The one that he will need to actually make phones at scale. That will cost billions ultimately, and the incumbents own the supply chain he needs to access. His effort is all but doomed.
3. OpenAI’s formula was easier for a startup to master. All they needed was money for the best AI engineers and scientists and money for GPUs, and they could create a blockbuster product. Rabbit needs the top engineers as well as extensive capital for manufacturing and distribution. There is a reason that hardware favors massive scale and a reason why hardware startups tend to focus on pinpoint innovations. The energy barrier is extreme.
These are three reasons why Rabbit is in an entirely different situation than OpenAI was circa 2022.
Both of those companies had it handed to them, like, literally got completely smoked by OpenAI, a company with a thousand employees +/- in San Francisco. The giants are incredibly vulnerable, just like the giants that Google and Apple disrupted such as IBM, Yahoo, AOL, etc.
I worked with one of the (many) teams at Microsoft who worked on Cortana.
The way the team leader explained it to me is that Cortana could do a lot more, but internal corporate politics prevented it. Rather than implementing the best solutions to user's problems, they had to do things like ensure Bing search handled certain results, to make sure that team stayed happy.
Or to take it to the extreme, if someone at Google came up with a device that directly beamed 100% correct search results into your brain, Google would never release the product because of the loss of search ad revenue.
By my reckoning, this approach is probably ten years ahead of its time. I simply cannot imagine that the tech is there right now to make this nearly as seamless as it will need to be to actually supplant the UI paradigms we have today, and it's going to take billions invested by the duopoly to get to that point.
But I really do think this is at least a fuzzy picture of where we're headed. Your iPhone in 2034 won't look like this, but you'll likely be able to trace some things back to it. There are a lot of pieces missing that we barely even know are missing yet, but it's incredibly exciting to see a startup try to jumpstart a step change like this.
Apple and Google presumably know about LLMs so I don’t really see this device being very influential.
I could imagine replacing my iPhone and my laptop with something Rabbit sized and a lightweight set of AR glasses.
But they need far more than the toy “OS” that Rabbit are demonstrating.
Perhaps at some point technology will allow glasses to have screens that are functional without making you look like, as they called them ten years ago, a Glasshole. Ten years may be enough time.
Anything that even threatens those business models will be strangled before it gets released.
IMHO Microsoft has the most to gain here. If they manage to turn Apple and Android into platforms that interface with MS made/ran/hosted LLMs, MS wins the day.
The problem is monetization, these LLM services are expensive, and telling users they need to pay another $30 a month to use their phone will be a hard sell. Hopefully LLM hosting costs go down, but if hosting LLMs becomes too affordable, MS's giant war chest of $ and hardware become less of an advantage.
The other danger is Apple/Google throwing a bunch of restrictions in place, but all the anti-trust scrutiny they are under right now could give them pause on attempting that.
i don't think apple or google are well-positioned to build this revolution, because they are too conservative and too bought-in on the old interaction model.
(if jobs were still around, different story. alas.)
they already have a habit of replicating the most useful apps natively
i don’t see why not here
replace siri once and for all, while running a local llm server for apps to tie into instead of an api to chatgpt
It will. I don't know which VC did the due diligence but the CEO once founded a scam company called RavenTech and the CTO just dropped out of CMU.
They used to be called CyberManufacture Co. and was selling NFTs. You see where this is going?
Even if Google builds something like this, gets market share and sells at a profit, it might still be a net loss for them because of all the advertising money and click tracking data they won't be getting.
Google definitely has the expertise to build this, but they also have an extremely risk-averse attitude resulting in layers upon layers of bureaucracy, and this product is literally cannibalizing their most important markets. I can't see this going over very well at all the internal reviews.
Apple was a lot leaner than Google when they build the iPhone (which was also cannibalizing one of their main products, the iPod), but they still had to make a completely independent team with no oversight except for Steve Jobs to get it right.
We also know how bad Google is at hardware. They seem to have gotten marginally better in recent years, but Pixels are still far from successful, even in countries where they are sold, which honestly isn't that many in the first place.
I'm very bearish on Google in this fight. Apple probably stands a much better chance, they've already proven they can cannibalize their own products, they need to do far less of that in the first place, as most of their revenue comes from hardware sales and subscriptions (which is perfect for something like this), they do have a lot of hardware expertise and they've already made major steps towards AR glasses, which might end up as a better form factor for such a device than a phone. Their pro-privacy attitude might be an impediment, this stuff works a lot better if you run it on a beefy GPU in the cloud instead of a tiny, battery-conserving chip in your phone, but I hope they find a good enough compromise.
ALL AI wearable companies should return their money to investors, and wait for the AR glasses by Apple.
AR glasses are the ultimate form factor. And as much as I hate monopoly, Apple has the right app/dev ecosystem, and will make Siri work.
The AR glasses race is between Apple and Meta Platforms.
This 'LLM-first phone' looks like a solution in search of a problem. But we'll see.
Has some CPP vibes.
Yes, google or apple will probably ship LLMs on their phones eventually and make an overall more compelling product than Rabbit _if_ you're only willing to have one device.
Considering the relatively low price (and assuming they actually deliver it in a timely fashion) I can see this being useful as a pocket assistant between now and whenever google et al eventually catch up, and potentially past then if rabbit keeps innovating.
[1] https://www.thelightphone.com/shop/products/light-phone-ii-b...
- the keynote video isn't scrollable behind the giant "share" button - the text on the keynote page is incredibly difficult to read (small text + very low contrast) - the homepage is full of dizzying animations that don't tell me anything about what the "phone" is actually supposed to do - "How it works" is just a set of animated cards. Once you figure out they're clickable (hoverable?), they spin around to reveal... half-formed ad copy? I wanted to know how it worked.
I'm willing to believe this is a real product only because of the "research" page, but it feels more like an aspiring web designer's portfolio piece. The flashiness of it all makes it really difficult to figure out what anything actually is.
While this thing does have a touchscreen keyboard it seems to me like the biggest selling point of the phone is actually something I don't want to use most of the time. I can't really see this succeeding for that reason.
Who actually interacts with a device like that? Is it an important problem enough to be paying $199 while I can do the same thing (often even better) with my phone?
https://arstechnica.com/gadgets/2022/11/amazon-alexa-is-a-co...
No. I could go deep into "rare language, weird slang, local dialect" hole, but simple "No" would be enough. Their website is exceedingly non user-friendly, too
Not sure I want to replace any other actions yet. Probably someday, but touch + apps is actually just incredibly good. A sub-1yo can start learning to use a smartphone / ipad, and a 6yo can put together an uber eats order. I'm not saying it can't be improved on, and I fully want a competent voice assistant in my phone. But it's gonna be real hard to do better than what we already have in our pockets on the things they're already good at.
So this presentation fell flat is for me are when it was talking about doing things pretty much everyone does just fine on their phone. But it was very exciting to imagine a ChatGPT+ device sitting around the house for anyone to pick up and interact with.