Wednesday January 10, 2024; 10:01 AM EST
- Learned about the rabbit r1, which is a new mobile device built on a Large Language Model. The underlying operating system, rabbit OS, does not run apps nor store files, its user interface is primarily natural language processing, and the device has a small touch screen to display information, including the text of verbal responses from the device. Conversation with rabbit is initiated by "push to talk" so it is not constantly listening. #
- Rabbit OS includes what the company calls a Large Action Model that knows about the APIs of applications like Spotify and Uber, which the operating system uses to execute actions. One asks Rabbit to play music and LAM uses the Spotify API to identify and play the music that fulfills the request. The demo shows how one can interact to play another song from the same album as the song currently playing or ask who has covered the song. Also demoed were requests for a ride, fulfilled via Uber, and order a pizza, fulfilled via Door Dash. Finally, the demo shows how one will be able to teach Rabbit how to execute certain actions that you want it to perform for you like generating an animation using Stable Diffusion.#
- What this company is doing here is using LLMs to enhance natural language processing so that it hears the speaker correctly, confirms what it heard, and then can perform specific actions. During the demo the person interacting with the device had an accent and did not appear to speak in any specific way and the device appeared to understand everything flawlessly. If this software and device works as reliably as demoed it addresses a frustration with all of the assistants from Google, Microsoft, Apple, and Amazon have in that none of these devices understand what is being spoken to it reliably. #
- The Rabbit R1 will have 4G LTE, WiFi, and Bluetooth connectivity and all AI processing is said to be on the device. One uses a portal to configure the interfaces with the particular applications one wants to use. It's not clear whether one can make phone calls using the device, and the presenter of the demo stated the device is not intended to replace a smartphone. I wonder what certifications has been obtained for the device, and securing them could impact the spring ship date.#
- The device is priced at $199 and there is no subscription model, which causes me to wonder how it is going to make money. The company has raised $30 million in total funding via two offerings. The company behind the device has nine people and is based in Los Angeles. Per the update from the second offering:#
Headquartered in Los Angeles, rabbit was founded by a group of top-ranking Kaggle Grandmasters, ex-Google engineers and repeat-entrepreneurs with extensive experience in creating AI hardware and operating high-performance computing clusters. A two-time Y Combinator alumnus, rabbit's Founder and CEO, Jesse Lyu, previously founded Raven Tech, a startup that pioneered conversational AI operatingsystems and was acquired by Baidu.
#- I don't know whether the device is going to sell, but I think the software is interesting and worth keeping an eye on to see how well it is received. If successful, the most likely outcome is for the company/software to be acquired to be used in a larger scale. The web site says that you can pre-order the device that will ship on Easter, 2024. #