Perplexity Launches Workflow AI
AI company Perplexity has unveiled “Perplexity Computer,” a multi-model AI system that can autonomously create and execute complex workflows. The technology is aimed at enterprises and developers, designed to automate processes that currently require human project managers.
The new system, dubbed a "general-purpose digital worker," moves beyond single-task automation to handle entire projects that could run for hours or even months. It operates by breaking down a user's stated goal into smaller tasks and then assigning these to specialized sub-agents that can conduct research, write code, design user interfaces, and manage deployments. Perplexity Computer orchestrates a suite of 19 different specialized AI models to accomplish these tasks. The system uses a core reasoning engine, Claude Opus 4.6, to delegate work to the best-suited model, such as Gemini for research-heavy tasks, Grok for speed, and Nano Banana for image generation. This model-agnostic approach allows the system to be updated as more powerful AI models become available. Real-world applications include automating the launch of a mobile app from research and design to coding and deployment. For business intelligence, it can perform competitor price comparisons, filter stocks for investment analysis, or build slide decks based on real-time data. The system interacts with software just as a human would, by reading the screen, using a file system, and operating a web browser in a secure, isolated environment. Founded in August 2022, Perplexity AI was created by Aravind Srinivas, Denis Yarats, Johnny Ho, and Andy Konwinski, who have backgrounds at Google, OpenAI, DeepMind, and Databricks. CEO Aravind Srinivas previously worked as a research scientist at OpenAI and Google, focusing on machine learning. The company's valuation has grown rapidly, reaching a reported $9 billion by the end of 2024.
