rabbit hints at future and partner hardware with 'rabbitOS intern' launch

Since launch, the company has been focused on agents, and this release hints at Android-like ambitions to be the OS of a broad AI consumer hardware ecosystem

People who I speak with often know I’m extremely bullish on this being the time for an exciting new period in the evolution of consumer hardware (and, at the same time, software and experiences). I’ve written before here about some of the recent entrants in the field focused specifically on building consumer devices from an AI-first perspective, including Humane (RIP) and Limitless – as well as rabbit, the company behind the r1.

I’ve also talked about how rabbit has generally improved the experience of r1 iteratively since its launch, despite stumbling out of the gate due to a mismatch between customer expectations around the product’s capabilities as described in pre-launch marketing materials and the reality once it was in their hands. Humane suffered from a similar affliction, but in its case the wounds were fatal – mostly because of the different approaches both companies took to getting their products to market.

Rabbit has also been very good about post-launch software iteration and development, as well as listening to, and engaging with their device owner and user community. They also landed some quick wins with sticky and charming features like ‘magic camera,’ which created pixel art-style renditions of photos taken with the device to be discovered and shared with a slight delay, almost like waiting for film to develop. And they also continued to work on their concept of the ‘Large Action Model’ or LAM, which was essentially an agentic AI system intended to perform tasks via the web on the user’s behalf before ‘agentic’ was the AI buzzword of the day.

In addition to the LAM, rabbit has continued to add new agentic experiences to its capabilities, including a way for users to teach it how to perform actions via ‘teach mode,’ and an Android agent that can effectively use Android apps on a user’s behalf. Now, it’s combining those capabilities and adding new ones to launch ‘rabbitOS intern,’ a new version of its operating system designed to capitalize on the clear appetite for consumer agentic products demonstrated by tools like OpenAI’s Operator and China’s Manus, while also charting a path forward for rabbit in terms of company evolution and revenue opportunities.

RabbitOS intern is aptly named for a couple of reasons, according to the company: First, it’s explicitly described as potentially “error-prone” and somewhat slow at launch. Second, it’s meant as a first step on an interactive path that should see it up-level its skills to eventually graduate to associate, mid-level, senior professional and beyond. It’s described as being able to perform tasks across “a variety of domains,” including doing this like creating a real estate research report for a specific area, making a fully functional (though simple) game in Three.js, creating a website for a TV show and coding a simple chiptune music sequencer as a web-based application.

Crucially, rabbitOS intern isn’t tied specifically to the company’s r1 hardware: At launch, it’ll be available as a free trial to all at rabbit’s ‘rabbithole’ website with some usage limits (with r1 owners getting more use). During the trial, the company will be evaluating pricing and plans for its eventual paid launch. The r1 launched as a product whose capabilities would specifically be available without an ongoing subscription, and those will remain with this as an optional upgrade once the trial period is over.

Rabbit says explicitly that rabbitOS intern is designed “to eventually power rabbit’s future products,” but also goes further to say it’s meant to also “integrate with any digital interface and other compatible devices,” which is why I suggested above that it feels like an Android-style approach to a future where mass market consumer experiences are primarily powered by AI under the hood.

I’ll caveat all of this by noting that I haven’t yet had the opportunity to try rabbitOS intern myself, and so can’t vouch for its state of ‘bakedness’ or capabilities. But what I like about it is what I’ve always liked about rabbit’s approach: It’s a novel and interesting, if very nascent, take on what the future could look like with entirely new device and software paradigms that use AI as the basis for their design and development, rather than just as a supplemental or additive feature.

I think that, as critics are extremely quick to suggest, it’s still much more likely that consumer adoption of AI will broadly happen via existing device and platform paradigms like iPhone, iOS and Android, used as the vehicle for AI software delivery both through OS integration and over-the-top through consumer apps. But while it’s less likely overall, I think it’s far more exciting – and potentially far more lucrative, from a market and investment perspective – to imagine and build towards a future where AI-first devices and software topple and supplant the relatively fixed and mature categories of devices like smartphones, laptops and computers.

As a company, rabbit is actively trying to create that future, and I think should be lauded for doing so, especially given that they launched their hardware with a relatively low selling point and no ongoing subscription commitment required, while iterating quickly and effectively post-launch. With rabbitOS intern, it’s debuting something that it admits is early, but that has the potential to become something big, and it seems to be doing so with even more care paid to not making its customers bear the brunt of the cost of testing and building something new in public.

If we get to a world where consumer AI devices and experiences are unshackled from our conventional, pre-AI devices and interaction models, it’ll be because of efforts like this to advance the state-of-the-art while also avoiding the alienation of people who don’t want to feel like they’re paying for the privilege of alpha testing the future.