Three Ways Codex Can Use a Computer

2026-06-16 · Source: Jason Liu · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Novice, medium

Summary

The article details three distinct methods for OpenAI's Codex to interact with a computer, extending its capabilities beyond mere coding to general computer use. These methods include "Computer Use" for native applications and cross-app workflows, "Chrome" for signed-in browser experiences and multiple tabs, and an "in-app browser" for public pages and visual review. "Computer Use" is the most powerful, enabling operations in apps like Spotify or finance tools, and can be enhanced by "Appshots" for context sharing and "Remote Control" for mobile steering. The "Chrome" integration leverages user profiles for tasks like music composition or social media monitoring, while the "in-app browser" facilitates visual annotation and context setting for other tools. The author emphasizes using the narrowest tool for a given task, providing a routing rule for efficient workflow management.

Key takeaway

For automation engineers building complex workflows, understanding Codex's distinct interaction methods is crucial. If your tasks involve native applications, use "@Computer"; for authenticated web experiences, "@Chrome" is ideal; and for visual review or public pages, "@Browser" is best. This structured approach allows you to design more robust and efficient automations, ensuring Codex can seamlessly navigate diverse digital environments without manual intervention, significantly expanding your automation scope.

Key insights

Codex can automate complex, multi-application workflows by integrating with native apps, Chrome, and an in-app browser.

Principles

• Use the narrowest tool that can finish the job.
• Goals define finish lines; automation manages waiting.
• Research and draft freely; ask before publishing.

Method

The article describes a routing rule for Codex tools: prefer plugins, use Appshot for context, @Browser for public pages, @Chrome for signed-in sites, and @Computer for native apps.

In practice

• Automate customer service interactions.
• Manage complex web forms and payments.
• Annotate web pages for collaborative feedback.

Topics

• Codex
• AI Automation
• Desktop Automation
• Browser Automation
• Workflow Orchestration
• Appshots

Best for: Automation Engineer, AI Engineer, Software Engineer

Related on AIssential

• Browser use with Codex plus GPT-5.6 and @Chrome is the best agentic workflow I've found — Combining Codex, GPT-5.6, and Chrome creates a powerful agentic workflow for advanced browser automation.
• Grab my Ultimate Guide to Codex and catch up to the 1 in 1,600 people using it every week (mostly no code!) — Codex transforms computer interaction by integrating deeply across systems, enabling automation of complex, multi-step jobs beyond basic chat.
• Computer and browser use in Codex (5 real examples) — AI desktop apps with browser extensions can automate complex digital tasks across various personal and professional domains.
• Codex Browser Use IS INSANE! Controls Your Computer & Automates Everything! — Codex, powered by GPT 5.5, offers autonomous AI agent capabilities for software development, testing, and computer control.
• Browser Agents vs API Automation: Which One Should You Use? — The optimal automation interface depends on the workflow's nature: APIs for structured records, browser agents for human-facing actions.
• Let Claude handle work in your browser — Claude for Chrome enables AI to directly interact with web pages for workflow automation.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Jason Liu.