Perplexity has introduced the “Computer”, a new tool that allows users to assign tasks and see them executed by a system that coordinates multiple agents running different models.
The company claims that the computer, which is currently available to Perplexity Max customers, is “a system that creates and executes an entire workflow” and is “capable of running for hours or months.”
The idea is that the user describes a specific outcome — something like “Plan and execute a local digital marketing campaign for my restaurant” or “Build an Android app for me that helps me do a specific type of research for my job.” The computer then considers sub-tasks and assigns them to as many agents as needed, running the models that Perplexity believes are best for those tasks.
The core reasoning engine currently runs Anthropic’s Cloud Opus 4.6, while Gemini is used for deep research, Nano Banana for image generation, VO 3.1 for video production, Grok is used for lighter tasks where speed is a consideration, and ChatGPT 5.2 is used for “long context recall and comprehensive search”.
This type of approach to the best model for the task differs from some competing products like Cloud Cowork, which only uses Anthropic’s model.
It all happens in the cloud with prebuilt integrations. Perplexity says, “Each task runs in a separate compute environment with access to a real file system, a real browser, and real tool integration.”
The idea is partly that this workflow was similar to what some power users were already doing, and the aim was to make it possible for a broader range of people who don’t want to deal with all that setup. People were already using multiple models and tailoring them to specific tasks based on perceived capabilities, while, for example, using MCP (Model Reference Protocol) to give those models access to data and applications on their local machines. Perplexity Computer takes a different approach, but the goal is the same: AI agents run specially chosen models to perform tasks related to your files, services, and applications.
Then there is OpenClause, which you can view as the immediate predecessor of this concept.
the story So Far
If you haven’t been following the wild OpenClaw craze, here’s a quick summary: Originally titled Cloudbot, then Moltbot, OpenClaw was an agentic AI tool that leveraged large language models to operate independently as a kind of background or ambient process on your local machine, performing a variety of tasks, from sorting your email history to building a website, to basically anything you can imagine.
<a href