Rabbit R1 and the Problem of Unfinished AI Products

The Rabbit R1 is a $200 AI-powered device, designed by Teenage Engineering, intended to operate as a portable, multimodal virtual assistant. The reviewer notes its close similarity to the Humane AI Pin, but identifies two intended differentiators: its lower price tag (no monthly subscription, unlike the Humane AI Pin's $700 cost and $24/month fee), and the planned "large action model" which aims to convert natural language commands directly into actions within apps—a departure from API reliance. The device features a physical scroll wheel and a swiveling camera, but its UI is quirky and often frustrating: navigation requires excessive scrolling, there's no back button, and the touchscreen is mostly unused except for terminal mode typing. Battery performance is poor (1,000 mAh, four-hour life, slow to charge), and many basic assistant features (alarms, timers, photo/video capture, calendar, email) are absent. The device frequently hallucinates responses, a persistent flaw in current AI assistants.

The "large action model" is meant to let Rabbit R1 interact with apps as a human user would (using UI elements like play/buy buttons), but as of review, only four apps (Spotify, Uber, DoorDash, Midjourney) are partially supported, often with errors. Rabbit claims 800 apps are "trained," but lacks a UI for most, and features like generative UI and teach mode (where Rabbit learns from the user's actions via mouse/keyboard) remain theoretical or "in the works". The R1's physical build reflects its price: bright orange plastic, basic camera and speaker, no accessories, and a 2.9" TFT screen with no auto-brightness.

The reviewer critiques a broader industry trend: releasing highly unfinished products and retroactively improving them, complicating the review process. Despite the promise of future capabilities, such as personalized assistants and teach mode, the reviewer urges: "Buy the product based on what it is today and not what it's promised to be in the future," noting that the low price may incentivize buyers to gamble on eventual improvements and provide training data, but without an independent appeal, Rabbit risks a "chicken and egg problem." The reviewer draws parallels with Tesla's early autopilot data collection and anticipates competing efforts from Apple and Google.