Rabbit's Second Act: From Hardware to Software with OS3

Revisiting Rabbit's Origins

Two years ago, Rabbit made headlines with its Rabbit R1, a device that promised to be the future of AI interaction through dedicated hardware. At CES 2024, it was heralded as an innovative leap forward in artificial intelligence. However, reviews were scathing; WIRED gave it a paltry 3/10 rating for its underwhelming performance.

"I don't think the Rabbit R1 was a flop," says Jesse Lyu, CEO and founder of Rabbit. "We've been through all of this before."

Lyu's words carry weight given the company's journey since then. The hardware, while ambitious in design, failed to deliver on its promises—a stark reminder that innovation alone doesn't guarantee adoption.

The Strategic Pivot to OS3

Now, Rabbit is launching OS3, a standalone operating system designed for cross-platform use. Unlike the R1, which required specific hardware, OS3 operates seamlessly across various platforms including desktop browsers, Telegram on phones, and even iMessage.

• Cross-Platform Compatibility: The system runs on cloud infrastructure but connects directly to local software, files, and web pages via a local agent node installed on Windows, Mac, or Linux.
• Universal Skill Integration: Users can integrate third-party AI agents simply by pasting a public URL into the chat interface—an effortless way to extend functionality without any technical expertise.
• No Subscription Required: OS3 follows a "Bring Your Own Key" (BYOK) model, allowing users to leverage existing API keys from providers like OpenAI or Anthropic—or link local open-source models.

Learning from Early Mistakes

After facing criticism and financial constraints post-R1 launch, Rabbit made tough choices. Rather than expanding the team or rushing into new hardware, they chose to focus on software improvements for the existing device. Over the past year and a half, more than 50 updates have been deployed, enhancing user experience and functionality.

The company also opened up the R1 hardware to community-driven enhancements, enabling integration with agents like OpenClaw or Hermes AI agents. This level of openness helped build a loyal community of over 12,000 users who provide valuable feedback for future developments.

Market Readiness and Competitive Landscape

The timing of OS3's release couldn't be better. The AI landscape has evolved significantly since the R1's launch. Today, agentic systems are all the rage, with major players like Meta pushing the boundaries of what AI assistants can do. However, this progress comes with challenges—including power consumption issues and regulatory concerns.

Meta's Muse agent recently faced backlash when Amazon banned it from using virtual machines to shop on behalf of users. This situation echoes the R1's controversial use of a VM for app execution, suggesting that some problems may be systemic rather than isolated incidents.

Financial Health and Future Vision

Despite facing financial strain and employee unrest due to delayed payments, Rabbit has secured additional funding and is now in a stronger position. Lyu claims the company has raised around $60 million and currently employs just 15 people—a lean team that's focused on efficiency.

Importantly, Rabbit has no plans for an R2. Instead, they're working on what Lyu describes as a Cyberdeck—a dedicated 'vibe coding' machine that will run OS3. This upcoming hardware is expected within months, further solidifying Rabbit's vision of bringing AI capabilities directly into users' workflows.

Real-World Applications and User Benefits

Lyu shares practical examples of how OS3 enhances productivity. For instance, he uses it to automate routine tasks like updating spreadsheets from weekly sales reports sent via email. By asking the agent through Telegram, OS3 completes the process without requiring manual intervention.

Another compelling example involves remote communication. During a video shoot, Lyu had his OS3 agent represent him on Slack, handling discussions with engineers and accessing files on his computer—all while he was unavailable. This demonstrates how OS3 can extend human presence in digital environments, making it more than just an assistant—it's a personal representative.

Privacy and Data Handling

Rabbit emphasizes privacy, stating that local agent nodes don't store, copy, or sell data. While conversations and memory logs remain on Rabbit's servers, file contents stay on the user's machine. When a task requires model reasoning, prompts pass through Rabbit's servers to the chosen provider, which processes them according to its own policies.

This approach strikes a balance between usability and privacy—offering powerful features while respecting user concerns about data security.

Conclusion: A New Chapter for Rabbit

Rabbit's path back into the spotlight with OS3 marks a significant evolution. Rather than clinging to hardware-centric solutions, the company has embraced a more flexible, software-first approach that aligns with current market realities. As AI becomes increasingly integrated into everyday workflows, tools like OS3 show how far we've come—and where we're heading.

With its blend of open-source integration, intuitive design, and strong privacy protections, Rabbit is well-positioned to carve out a niche in the rapidly growing agentic software space. Whether this marks the beginning of a new era for AI interaction remains to be seen—but one thing is certain: the future belongs to those who adapt.