Rabbit R1 source code [part 1]

Original Article ↗ Hacker News Discussion ↗

Leaked code from the Rabbit R1 “AI assistant” suggests its much-hyped “Large Action Model” may largely be a set of Playwright automation scripts driving a few web apps, sparking claims of overblown marketing and even accusations of a cash grab similar to the Humane AI Pin. Commenters debate whether this kind of smoke‑and‑mirrors demo is standard tech hype or crosses into scam territory, especially given the product’s reliance on users’ credentials and opaque cloud infrastructure. Others argue the underlying idea—voice-driven automation on top of existing apps—is conceptually interesting but better suited to smartphones or open, hackable hardware, and note that platform lock‑in on iOS and Android makes it hard for third parties to build trustworthy, deep assistants.

Perception of Rabbit R1 and Humane Pin

  • Many see Rabbit R1 (and Humane Pin) as hype-driven or “cash grab” products rushed into market before iOS/Android integrate similar AI features.
  • Others argue Humane at least had years of serious hardware R&D and only pivoted to LLMs when they became unavoidable, calling “cash grab” unfair but admitting execution is poor and vision unclear.

Demo Authenticity and “Scam vs MVP”

  • The Rabbit keynote is heavily criticized as misleading: “live” demos (e.g., trip booking, candidate filtering) appear staged and in at least one case visibly fail on screen.
  • Some note that big tech frequently fakes or tightly scripts demos; the debate is whether Rabbit crossed into deception.
  • A common view: if a shipped product behaves far below the demo with no clear disclaimer, it’s closer to a scam than an MVP.

Source Code Leak and “Large Action Model”

  • Leaked code (via zip archives on file hosts, then briefly mirrored on GitHub) appears to show a Node.js backend orchestrating Playwright-controlled browsers/VMs.
  • Critics claim this disproves marketing around a “Large Action Model” and shows it is “just Playwright scripts.”
  • Others counter that an LLM-based decision layer could still exist elsewhere; the leak seems to cover only browser “minions,” not the core model. Even Rabbit’s CTO (via Discord screenshot) claims LAM “lives elsewhere.”
  • No clear consensus: the leak exposes infrastructure, but whether the “LAM” is fundamentally novel remains unclear.

Security, Credentials, and Distribution Concerns

  • Serious concern about Rabbit’s approach of logging into third-party sites on users’ behalf (Midjourney, Spotify, etc.), potentially via session hijacking rather than APIs/OAuth.
  • Some compare this to normal trust in platforms (Android/iOS, OAuth, password managers); others distinguish between app sandboxing vs typing raw credentials into an opaque remote automation system.
  • Secondary debate about distributing leaked source as zips: some worry about normalizing risky download behavior and malware; others consider archives standard and useful for hashing and mirroring.

Platform Lock-In and Assistants

  • Several argue that meaningful AI assistants are blocked by mobile OS lock-in: third parties lack APIs to send texts, place calls, control apps, etc.
  • Some call for antitrust action on private APIs and insist users should be able to swap the “agent” on their device.
  • Others note Apple’s existing SiriKit but say Siri’s recognition and UX lag behind newer systems (e.g., Whisper-like performance).

Form Factor, Use Cases, and Alternatives

  • Many think the R1’s core “assistant” functionality belongs on phones, watches, or earbuds, not a new dedicated gadget you must carry in addition.
  • There’s extensive skepticism that voice-first devices will replace screens for complex tasks (flights, restaurant search), though some envision a future of unobtrusive voice-driven wearables and AR glasses.
  • A subset of commenters bought or would buy R1 mainly as a hackable Mediatek-based toy; they wish Rabbit/Humane would embrace openness and self-hosted backends.
  • Open-source or alternative projects (OpenAdapt, openinterpreter, hackable glasses, etc.) are cited as more interesting paths for experimentation.

Teenage Engineering and Hardware Design

  • Rabbit’s industrial design is widely praised, and Teenage Engineering’s role is dissected.
  • Some argue TE risks reputational damage by associating with hype products; others think it’s just contract design work and their niche audience won’t care.
  • This spills into a tangential but heated debate over TE’s pricing, build quality, and alternatives in the music hardware world.