Perplexity's Objective-Driven AI Operating System

Perplexity has introduced an AI operating system that executes objectives rather than just following instructions. This persistent digital proxy has local access to your files and can be controlled remotely from any device. Safety is ensured through user approvals, activity logs, and a dedicated kill switch.

Key Points

• AI operating systems focus on high-level objectives instead of step-by-step instructions.
• The system acts as a persistent digital proxy with local access to files and apps.
• Users maintain control through approval requirements, logs, and a kill switch.
• The platform is accessible from any device, anywhere.

Sentiment

HN is strongly negative and dismissive of this product. The community broadly agrees that the productivity claims are fabricated, the product is derivative of open-source work, the privacy guarantees are contradictory, and the entire announcement represents peak AI hype theater rather than genuine innovation.

In Agreement

• If Perplexity ships this successfully, it could normalize always-on AI agents with OS-level access and push the whole space forward.
• AI agents could theoretically eliminate large-scale organizational coordination overhead in ways analogous to how computers replaced rooms of human calculators.
• Some users were open to trying it on a secondary, non-primary machine—suggesting the concept itself has cautious appeal.

Opposed

• The productivity claims ($1.6M saved, 3.25 years of work in four weeks) are mathematically dubious and rely on self-reported survey methodology that systematically inflates savings.
• The product is essentially a commercial repackaging of the open-source OpenClaw project, with several users alleging Perplexity even submitted low-quality 'slop PRs' to that project.
• Despite claiming to offer local, private access, the system connects to Perplexity's cloud servers—making the privacy and security claims contradictory.
• The landing page and product description are so vague and confusing that many users couldn't understand what the product actually does.
• Perplexity has no competitive moat if it relies entirely on third-party AI APIs from OpenAI, Google, or Anthropic.
• An always-on AI agent with full OS-level access represents a serious security risk, and the 'kill switch' feature is seen as an implicit acknowledgment of this.
• The product exemplifies peak AI hype—a generic LLM wrapper applying technology to problems rather than solving real user needs.