Rabbit Debuts ‘Large Action Model’ Agent on r1 Platform Starting October 1!

  • Editor
  • September 24, 2024
    Updated
rabbit-debuts-large-action-model-agent-on-r1-platform-starting-october-1

Key Takeaways:

  • Release Date: The web-based version of Rabbit’s Large Action Model (LAM) will be available on October 1 for r1 devices, marking a major update since the product’s initial launch in early 2024.
  • Capabilities: The new LAM is designed as a general-purpose web agent, enabling the r1 to interact with any website and perform tasks like buying tickets or registering domains.
  • Privacy Measures: The agent operates through a cloud-based browser, with plans for local versions like a Chrome extension to enhance user privacy without accessing full credentials.
  • Development Stage: The LAM is still a “playground version”, functional but in need of further refinement, with prompt engineering required for optimal performance.

Rabbit is set to release a major update on October 1 that will bring a new version of its Large Action Model (LAM) to the web. The r1 device, which debuted earlier in 2024, initially garnered interest as a highly anticipated gadget but faced setbacks as it failed to deliver on many of its early promises.

Rabbit’s CEO Jesse Lyu admitted that the company had set overly ambitious expectations at launch. With the release of the LAM update, Rabbit aims to transform the r1 into a versatile, web-based agent capable of performing various tasks across any website.

Previously, when the company launched the Large Action Model (LAM), this is what users’ reaction was!

Comment
byu/jd_3d from discussion
inLocalLLaMA

Since its initial release, Rabbit has focused on 16 over-the-air updates, including bug fixes, response time improvements, and minor feature additions.

Despite these efforts, the r1 remained limited in scope, only interacting with specific services like Uber and Spotify. Lyu explained that the early version of the LAM was based on recordings collected from data laborers and was not a true general-purpose model.


The new web-based LAM is designed to be a generic, web-based agent that handles a wide range of tasks, such as buying tickets, registering domains, and playing online games.

Lyu demonstrated the agent’s potential by having it perform tasks like registering a domain for a film festival and purchasing an r1 device, initially opting for listings on eBay before being directed to buy from the official website.

Comment
byu/jd_3d from discussion
inLocalLLaMA

The LAM operates through a clean, cloud-based browser that ensures each session remains secure and isolated.

Rabbit plans to develop local versions, including a Chrome extension, allowing users to work within their existing sessions.

The current model does not have access to user credentials, addressing privacy concerns, and Rabbit is considering a walled-off small language model for secure logins in future versions.


While the LAM shows potential, it remains a “playground version,” meaning it is functional but not fully refined.

Prompt engineering is often required, as how requests are phrased greatly affects the outcome. Rabbit plans to introduce a “teach mode” that would allow users to guide the agent through specific tasks, enhancing its adaptability and personalization.


Rabbit is also exploring a desktop version of the LAM that can interact with software like word processors and music players. This desktop agent is in the early stages but has shown promise, managing tasks simply by interacting with software interfaces.

However, the LAM’s broader appeal remains challenging due to the absence of a clear, standout application demonstrating its practical value.


Lyu acknowledged that while the agent is impressive, its practical uses are not always obvious, particularly for users who already spend extensive time on traditional browsers.

Rabbit’s decision to develop the LAM as a standalone web agent rather than a mobile app reflects a strategic effort to avoid direct competition with tech giants like Apple and Google.


Lyu explained that creating an app would likely lead to conflicts with these companies, restricting third-party AIs from outperforming native assistants like Siri.

Additionally, launching as an app would have subjected Rabbit to substantial revenue cuts due to app store fees.

The LAM’s development as a third-party AI capable of interacting with various services aligns with Rabbit’s vision of a cross-platform, generic agent system.


Lyu mentioned that the company might explore a dedicated Rabbit phone in the future despite initially focusing on a simpler device.

As Rabbit prepares to roll out the new LAM update, the company is setting realistic expectations. While the update represents progress, it has limitations, and much room remains for improvement.


The LAM update will be available to all r1 owners this week, with instructions on how to utilize the new features expected soon. Rabbit hopes this update will help restore confidence in the r1 and showcase the LAM’s potential as a versatile tool for everyday tasks.

Lyu concluded with a measured statement, noting, “We’re setting the expectations right. It’s not perfect. It’s just the best the human race has achieved so far.”

This update marks a critical moment for Rabbit as it works to fulfill its early promises, refine its technology, and demonstrate the LAM’s practical value to users.

For more news and trends, visit AI News on our website.

Was this article helpful?
YesNo
Generic placeholder image

Dave Andre

Editor

Digital marketing enthusiast by day, nature wanderer by dusk. Dave Andre blends two decades of AI and SaaS expertise into impactful strategies for SMEs. His weekends? Lost in books on tech trends and rejuvenating on scenic trails.

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *