Perplexity Launches 'Computer' AI Agent
AI search company Perplexity has launched "Perplexity Computer," a new multi-model AI system. The platform is designed to autonomously create and execute entire workflows, signaling a move toward more agentic AI that can handle complex, multi-step tasks for enterprise users.
Perplexity's "Computer" moves beyond single-model systems by orchestrating 19 specialized AI models to execute tasks. The system uses specific models for different jobs, such as Opus 4.6 for core reasoning, Gemini for in-depth research, and others for generating images and video. This multi-model approach allows it to break down a high-level goal into subtasks and assign the best-suited AI to each part of the workflow. Unlike AI agents that run on a user's local machine, Perplexity Computer operates in a controlled cloud environment. This architecture is designed to reduce the security risks associated with AI agents accessing local files and systems. Each task runs in an isolated environment with its own file system and browser access, creating a secure "harness" for the AI's operations. The system is designed to be an "always-on" digital worker, capable of running complex workflows for extended periods without continuous user input. It can perform tasks like monitoring competitor pricing, analyzing financial data, creating presentations, and even building applications from a single prompt. When it encounters a problem, it can autonomously create sub-agents to find solutions, such as researching missing information or finding API keys. The founding team of Perplexity AI, established in August 2022, includes CEO Aravind Srinivas, Denis Yarats, Johnny Ho, and Andy Konwinski. Srinivas, an IIT Madras graduate with a Ph.D. from UC Berkeley, previously held research positions at OpenAI, Google Brain, and DeepMind. The company has attracted significant investment from notable backers including Jeff Bezos and NVIDIA. In the enterprise space, companies like Volvo Cars US, Monogram Health, and Stripe are using Perplexity's technology. Use cases range from market analysis and streamlining operations to enhancing customer engagement with intelligent chatbots. For example, law firm Gunderson Dettmer and the finance company Rho are using the enterprise version to speed up research. Compared to other agentic AI systems like OpenAI's Operator, Perplexity Computer is positioned as a more accessible option, integrated into its Pro subscription. While some agents may offer higher autonomy with local control, Perplexity's cloud-based approach prioritizes safety and ease of use for complex, research-heavy workflows.
