Perplexity Launches 'Computer' AI Workflow System
AI search company Perplexity has launched "Computer," a new multi-model AI system. The platform is designed to create and execute entire complex workflows autonomously. This could allow companies to automate processes from data analysis to customer service with minimal human intervention.
The "Computer" system is not a single AI model but an orchestrator for 19 specialized models. It uses Anthropic's Claude Opus 4.6 for core reasoning and coding, Google's Gemini for deep research, and Grok for faster, lightweight tasks, assigning the best model for each specific job. Launched on February 25, 2026, the system functions as a "digital employee" by breaking down a high-level goal into a series of subtasks. It then creates and deploys autonomous sub-agents that can work in parallel on things like web research, data processing, and coding within a secure cloud environment. This platform represents a strategic shift for Perplexity, moving the company from a "knowledge engine" focused on answering questions to an "action engine" designed to deliver completed projects. The goal is to translate user outcomes directly into finished tasks, handling multi-step workflows that normally require a human moving between different applications. Access to "Computer" is initially limited to subscribers of the $200-per-month "Perplexity Max" tier. The service uses a credit-based system, reflecting the high computational cost of running nearly 20 models simultaneously and marking a move away from the "all-you-can-eat" subscription model common in the industry. Founded in 2022 by Aravind Srinivas, Denis Yarats, Johnny Ho, and Andy Konwinski, Perplexity AI has raised over $1.2 billion in funding. The company is betting its $20 billion valuation (as of Sept. 2025) on this pivot to execution, with backing from investors like Jeff Bezos and Nvidia.
