Back to blog

June 7, 2026 · 4 min read

Computer use comes to your machine and the cloud

TaskForceAI computer use now runs both locally via the desktop app and in cloud Linux desktops, with a live Computer Theater view of agent actions.

Local computer use that runs on your own machine through the desktop app

Virtual computer use on a cloud Linux desktop for isolated runs

Computer Theater shows a live, annotated feed of every agent action

Two ways to give an agent a computer

Some tasks need more than tool calls. They need an agent that can actually drive a computer: open apps, click around, and work through a real interface. This week computer use arrived in two forms.

Locally, the desktop app lets an agent operate your own machine. Virtually, TaskForceAI can spin up a cloud Linux desktop so the agent works in an isolated environment instead of yours. Same capability, different blast radius, and you choose which fits the task.

Watch the work happen

Computer use is only trustworthy if you can see it. The Computer Theater view expands automatically when an agent starts working, streaming a live feed of the desktop with clicks and cursor movement drawn directly on screen.

  • Local execution on your machine via the desktop app
  • Virtual cloud Linux desktops for isolated, disposable runs
  • Live annotated feed with an auto-expanding theater and manual controls

Built to be observed, not hidden

The whole point is a visible agent. Whether it runs locally or in the cloud, computer use keeps the actions on screen so you can follow along, step in, or stop the run when you need to.

Build with TaskForceAI

Explore the docs, launch the web app, or install a native surface for deeper agentic workflows.