wouldit.work

AI tools

Rabbit R1: 100,000 preorders for an AI gadget that reviewers said barely worked

Rabbit's $199 R1 sold on a CES keynote promising a 'Large Action Model' that would use apps for you. Reviews found most of it didn't work, and its interface turned out to run as an Android app.

Product
Rabbit R1
Founder
Jesse Lyu (founder and CEO)
Launched
2024
Model
One-time hardware purchase

Published · 2 min read · 3 sources

Key takeaways

  • More than 100,000 people preordered the $199 R1 based on CES promises and demos, The Verge reported.
  • The Verge's review found the headline 'Large Action Model' features mostly didn't work; its launcher ran on a Pixel 6a as an Android app.
  • Rabbit's CEO later admitted 'a big gap between the high expectations of our users and the original software experience we delivered'.

What it is

The R1 is a $199 handheld AI device with a push-to-talk button, screen, camera and scroll wheel. Its pitch was a "Large Action Model" that learns apps and uses them on your behalf (The Verge review).

Timeline

  • CES 2024: debuted with a keynote "filled with big promises and impressive demos"; more than 100,000 people preordered (The Verge review).
  • 30 Apr 2024: Android Authority ran Rabbit's launcher APK on a Pixel 6a. CEO Jesse Lyu said "rabbit r1 is not an Android app" and pointed to cloud services and firmware changes (The Verge).
  • 2 May 2024: The Verge's review called it "underwhelming, underpowered, and undercooked"; only four app integrations existed and Uber and DoorDash never completed a task in testing (The Verge review).
  • Later: Rabbit shipped rabbitOS 2, a full redesign, after "more than 30 OTAs over the past 16 months" (rabbit).

Why it worked (as a launch)

  • A cheap, playful object. At $199 it was cheaper than the Humane AI Pin, and the reviewer liked the orange design and push-to-talk button (The Verge review).
  • A big demo at the right moment. The CES keynote sold a vision of an assistant that does things, not just answers questions (The Verge review).
  • A few genuinely useful bits. The reviewer's favourite feature was the voice recorder with AI summaries (The Verge review).

What to learn

  1. Preorders validate a demo, not a product. The gap between them is where trust is lost.
  2. If your product could be an app, assume people will ask why it isn't one (The Verge).
  3. Don't sell agent features before they work end to end. A failed order is worse than no feature.
  4. Shipping updates after launch helps, but reviews written at launch last much longer.

Risks and what didn't work

  • Basic tasks failed: it misidentified objects, mixed up Spotify requests and drained its battery quickly in early testing (The Verge review).
  • Logging in to third-party services through Rabbit's virtual browser "feels icky and insecure", the reviewer wrote (The Verge review).
  • It competes with the smartphone people already carry, which is getting AI features too (The Verge review).
  • Rabbit's own CEO acknowledged the original software missed users' expectations (rabbit).

FAQ

Short answers drawn from the sections above.

What is Rabbit R1?

The R1 is a $199 handheld AI device with a push-to-talk button, screen, camera and scroll wheel. Its pitch was a "Large Action Model" that learns apps and uses them on your behalf (The Verge review).

Why did Rabbit R1 work?

  • A cheap, playful object. At $199 it was cheaper than the Humane AI Pin, and the reviewer liked the orange design and push-to-talk button (The Verge review).
  • A big demo at the right moment. The CES keynote sold a vision of an assistant that does things, not just answers questions (The Verge review).
  • A few genuinely useful bits. The reviewer's favourite feature was the voice recorder with AI summaries (The Verge review).

What can indie builders learn from Rabbit R1?

  1. Preorders validate a demo, not a product. The gap between them is where trust is lost.
  2. If your product could be an app, assume people will ask why it isn't one (The Verge).
  3. Don't sell agent features before they work end to end. A failed order is worse than no feature.
  4. Shipping updates after launch helps, but reviews written at launch last much longer.

What didn't work for Rabbit R1, and what are the risks?

  • Basic tasks failed: it misidentified objects, mixed up Spotify requests and drained its battery quickly in early testing (The Verge review).
  • Logging in to third-party services through Rabbit's virtual browser "feels icky and insecure", the reviewer wrote (The Verge review).
  • It competes with the smartphone people already carry, which is getting AI features too (The Verge review).
  • Rabbit's own CEO acknowledged the original software missed users' expectations (rabbit).

Sources

Every link cited above, in order of first use.

  1. 1.The Verge review (theverge.com)
  2. 2.The Verge (theverge.com)
  3. 3.rabbit (rabbit.tech)

Related case studies

More in AI tools →