Perplexity's 'Computer' AI Agent Takes Control, Orchestrating Other Models and Prioritizing Your Security

Perplexity's 'Computer': Your AI Agent Command Center

You’re now able to command a fleet of AI models with Perplexity’s new “Computer.” This isn’t just another chatbot; it’s a system designed to orchestrate other powerful AIs – think Anthropic’s Claude Opus 4.6, Gemini, Veo 3.1, and even ChatGPT 5.2 – giving you a central hub for complex tasks. Perplexity positions Computer as a way to run AI workflows for hours, even months, without interruption.

The core of Computer relies on the Model Context Protocol (MCP), a technology gaining traction in the world of advanced AI agents. This allows different AI models to communicate and share information, creating a more cohesive and powerful workflow. Perplexity is aiming to give you a more streamlined and efficient way to leverage the best of what each AI has to offer.

Perplexity emphasizes that Computer is built with security in mind. This is a direct response to concerns raised by earlier, less controlled AI agent systems. You can expect a more contained and secure environment for your AI workflows.

The Lessons Learned From OpenClaw: Power and Responsibility

Before Perplexity’s Computer, there was OpenClaw. This earlier AI agent demonstrated incredible power, but also exposed significant security vulnerabilities. OpenClaw utilized systems like ClawdBot and Moltbot, and also leveraged the Model Context Protocol (MCP) to push the boundaries of AI capabilities.

OpenClaw could create, modify, and even delete files on your system, extending its influence “far beyond what most users could achieve with existing models and MCP.” This observation came directly from OpenAI CEO Sam Altman, highlighting both the potential and the danger of such a powerful agent. Specifically, files like USER.MD, MEMORY.MD, SOUL.MD, and HEARTBEAT.MD were all within OpenClaw’s reach.

The vulnerabilities of OpenClaw – including prompt injection and unauthorized file deletion – underscored the need for more robust security measures. These flaws allowed the agent to make changes without your explicit approval, raising serious concerns about data security and control. Perplexity’s Computer is, in part, a response to these risks, aiming to provide you with a more secure and reliable agentic AI experience.

What This Means For You

Agentic AI is changing how you interact with technology. Perplexity’s Computer offers a glimpse into a future where AI manages other AIs to achieve your goals, automating complex workflows over extended periods. Imagine automating research, content creation, or data analysis with a team of specialized AI agents working under your direction.

However, the story of OpenClaw is a crucial reminder: immense AI power demands robust security. You need to be confident that your data is protected and that the AI is operating within defined parameters. Expect ongoing innovation focused on balancing power and security as agentic AI becomes more commonplace.

For now, Perplexity’s Computer is available to Perplexity Pro subscribers. This means you’ll need a paid subscription to access this advanced functionality. The cost of Perplexity Pro is $20 per month, or $200 annually. This gives you access to the full suite of Perplexity’s features, including the new Computer agent orchestration capabilities.

The Bottom Line

Perplexity’s “Computer” is a significant step toward more powerful and automated AI workflows. It learns from the past, addressing the security concerns raised by earlier systems like OpenClaw. While the technology is still evolving, it offers you a compelling vision of a future where AI agents work seamlessly together to tackle your most complex tasks. If you’re a Perplexity Pro subscriber, it’s worth exploring to see how it can streamline your work and unlock new possibilities.