quarterly update: best-selling AI hardware, LAM 1.5, AI-native desktop OS, and more
The first few months of 2024 have been a wild ride. Since we announced r1 in January, we’ve made big improvements to the speed, accuracy and functionality of r1, to the point where it’s the best-selling AI device, having sold more than 100,000 units. This is a humbling achievement that was only possible with the support of our investors, partners, and community, which has quickly grown to become one of the most active and engaged communities in the industry.
Now that r1 is out in the real world, we wanted to share a recap of how far we’ve come in this short time, as well as what’s to come in the next stage for rabbit.
First r1 OTA Updates
First things first, we’ve already been getting a lot of very valuable feedback from the first group of users who have begun receiving their r1 devices. This includes hearing the feedback both from our active community as well as media reviews, which we take very seriously and are actively addressing. This week, we released our first two over-the-air updates where we made the following improvements:
On r1 / rabbit OS, users are seeing idle battery performance improving by up to 5x. We added time zone selection to r1’s settings menu, improved time zone updates based on GPS location and improved GPS location services with enhanced AGPS. rabbit OS is now more aware of when to use GPS to query your location and how to leverage that information. We have also improved the accuracy regarding time-related information during conversation. We improved the music playback UI on r1 and are proactively detecting server connections and releasing linked resources. We improved reliability for Bluetooth connectivity and will keep optimizing it. Users will find the meeting assistant / recorder is more stable and can support longer durations. We also improved bi-directional translation and weather / time related experiences, and have further reduced LLM hallucinations. On rabbithole, we improved the loading UI, error messages when your session expires, and QR code readability. We also made the TTS error messages from spreadsheet queries more informative.
Bringing our users quick updates is a top priority for us. Stay tuned to our social channels where we will share the release notes for future updates.
r1 Pickup Party - A Community Celebration
To celebrate the first r1 deliveries, we held our first community gathering - the r1 pickup party - last week at the iconic TWA Hotel in New York City. The vibe was very retrofuturistic, and the atmosphere was buzzing. More than 300 customers joined us to be among the very first to pick up their orders in person. As part of a live-streamed keynote, we conducted the world’s first r1 unboxing and did the first live demos of rabbit OS, showing everything from using rabbit eye vision to edit spreadsheets to ordering food on DoorDash. We put r1 through its paces at the event, and we’re looking forward to hearing what our users have to say as they start using the device more on a daily basis. You can watch the full keynote at https://www.rabbit.tech/live-unboxing.
New Partners, New Investors and New Additions to the rabbit Team
rabbit continues to move very fast in our mission to build out this new AI hardware space. In the past 90 days, we’ve quadrupled the size of our engineering team and have continued adding tremendous value to r1 with some of the best partners in AI, including Perplexity, ElevenLabs, OpenAI and Anthropic. In January, we announced that we were teaming up with Perplexity to provide real-time, precise search results on r1, without the need for a subscription. In March, we announced that we would use ElevenLabs’ ultra high-quality, low-latency voice models to give voice interaction with r1 a more natural feel. We also recently closed another funding round led by Sound Ventures.
At the pickup party, we announced an exciting new addition to the rabbit team: Jesper Kouthoofd, founder and CEO of teenage engineering, is bringing his talents and experience to rabbit as Chief Design Officer. Jesse and Jesper first met more than 10 years ago and immediately connected over their shared passion for synthesizers, music and product design. Now their collaboration will become even deeper, with Jesper working with Jesse to guide the future product design for rabbit. Jesper will maintain his roles of head of design / CEO at teenage engineering, and we’re already getting to work on the next generation of rabbit products with him.
LAM 1.5 - Bridging Digital Interfaces with the Physical World
We’ve made significant progress with our Large Action Model (LAM) on digital interfaces over the past few months. Now, we’re looking ahead to the next stage of possibilities with LAM 1.5, which will bridge the gap between the digital and physical world. Building on what LAM 1.0 has learned, LAM 1.5 aims to understand physical objects, associate them with their digital counterparts, and operate those products for you. For example, imagine pointing to your home’s smart thermostat and asking your device to change the room temperature for you. LAM 1.5 aims to do just that by identifying the right make and model of the device and operating it on your behalf.
Generative UI - Instantaneous Design Automation
As LAM continues learning more apps at a rapid pace and rabbit OS becomes more familiar with the person using the device, its interface will also become more personalized, dynamic, and proactive. We are exploring a very new direction to keep up with that pace and bring more supported apps to our users more quickly. With generative UI, we aim to build a system that’s truly personal and adapted to the end user with a tool that converts the great ideas of our designers into a final product in no time. Once an app is supported, the UI will be ready right away.
Personalized Computing - From Mobile to Desktop OS
Finally, we’re working on an upgraded version of rabbit OS that will convert a single website today, rabbithole, to a full AI-native desktop operating system. We are currently building the bare bones of rabbit OS, first ensuring that your services are connected and your logs and journals are gathered together in the same place. The next step will be to build rabbithole - the brain of your personal AI assistant that remembers your preferences and needs in order to serve you better - to a point where it can handle much more intensive jobs and requests through a desktop OS.
As we’ve said before, our mission is to create the simplest computer that you don’t need to learn how to use it. We believe that in the future, users will have the power to use natural language to get results. Thanks to all of you who are taking this journey with us.