Rabbit's r1 Enhances Chat and Timer Features, Yet Its App-Based 'Action Model' Remains Absent

Rabbit's r1, the AI assistant gadget that generated considerable buzz at CES, has released some updates. However, these enhancements are unlikely to sway its critics just yet.

A new “beta rabbit” mode introduces advanced conversational AI capabilities, particularly for handling complex or multi-step instructions. It now offers improved follow-up questioning when it requires clarification on certain tasks.

For instance, a user might ask:

“Beta rabbit, can you suggest three books similar to 'The Power of Now,' including page length, year of release, and ratings? Please save that as a note titled 'Reading List' and include pictures of the authors.”

Followed by: “Beta rabbit, can you also provide summaries of those three books?”

Other suggested uses include creating travel itineraries or finding product recommendations. As casual users of various chatbots, we recognize that while these tasks can showcase the technology's potential, they often lack real-world practicality. Chatbot-generated itineraries can be erratic, and comparing specs and prices can be cumbersome on a small device. Trusting sketchy book recommendations is also a tall order.

Additionally, there are improvements in alarm and timer functionalities (you can view all the new features here). While these updates are welcome, some are met with a “wait a second…” response. For example, a user may say, “Set a timer for baking chocolate chip cookies.” But doesn’t the AI need specifics like temperature, quantity, and type? Without that context, the outcome could easily be disastrous. However, asking, “How long should two dozen chocolate chip cookies bake at 300 degrees?” could yield useful advice.

What everyone is eagerly anticipating is the “large action model” (LAM) the company promoted back in January. Initially seen as an ambitious yet attainable objective, the model is supposed to autonomously navigate phone and web app interfaces to execute user-defined tasks. However, this capability has yet to be demonstrated outside of promotional videos, and claimed functionalities are indistinguishable from standard API calls or action scripting.

Despite limited opportunities to use the r1 since receiving it for review, I remain hopeful about the gadget’s future potential. That’s why I haven’t tucked it away in a drawer—yet.

I’ve reached out to Rabbit for updates regarding the LAM and will provide information if I receive a response.