Back to stories
Tools

OpenAI Turns Codex Into a Super App With Computer Use, Atlas Browser, and Image Generation

Michael Ouroumis3 min read
OpenAI Turns Codex Into a Super App With Computer Use, Atlas Browser, and Image Generation

OpenAI has pushed Codex well past its roots as a coding assistant. In an update rolled out on April 16, 2026 and expanded through the week, the company's desktop app now drives other applications on your computer, runs its own in-app browser, generates images, remembers context across sessions, and supports more than 90 new plugins. The company is openly calling it a super app — an ambition it's building in public rather than announcing as a finished product.

Head of Codex Thibault Sottiaux summed up the direction plainly: "We're building the Super App in the open and evolving it out of the Codex app." With roughly 3 million weekly users and around a million more signing on each month, the update reframes Codex as OpenAI's bid to own the autonomous-agent layer on developer and knowledge-worker desktops.

Computer use comes to the desktop

The headline feature is background computer use, available on macOS at launch (excluding the EU, UK, and Switzerland for now). Codex can now see, click, and type on the operating system with its own cursor, and multiple agents can operate in parallel while the user keeps working. OpenAI says it has been using the same capability internally for Codex's own quality assurance — a tell that reliability matters to the team more than flashy demos.

Paired with that is a new in-app browser built around OpenAI's Atlas. It currently supports opening local and public pages and letting users leave page-level comments that Codex can then act on. Full browsing and analysis features are expected to follow in later previews.

Images, memory, and scheduled automations

Codex can now generate and refine images using gpt-image-1.5 directly inside the app, at no additional cost — useful for stitching screenshots, design mockups, or UI references into a working session. A new memory system carries over preferences and corrections between sessions, and heartbeat automations let Codex schedule future prompts, monitor long-running tasks, and resume work across multiple days.

The plugin expansion is equally broad: JIRA, CircleCI, GitLab, and Microsoft's productivity suite are among the 90-plus new integrations, each combining skills, app connections, and MCP servers to widen the agent's reach across enterprise toolchains.

Why it matters

The practical implication is that OpenAI is stacking agent capabilities in one desktop surface rather than distributing them across a family of standalone tools. With GitHub review handling, multi-tab terminals, remote SSH in alpha, and a file sidebar for PDFs and documents, Codex is positioning itself as the default AI layer in front of a developer's workflow — not a sidekick opened when something breaks.

Personalization features — along with computer use in the EU and UK — are not yet available for Enterprise, Edu, EU, and UK accounts, with OpenAI saying those rollouts will follow soon, and some capabilities remain in preview. But the shape of the product is now clear: Codex is where OpenAI wants autonomous desktop work to live, and the pace of its rollout suggests competitors will need their own super-app answer quickly.

Learn AI for Free — FreeAcademy.ai

Take "Prompt Engineering Practice" — a free course with certificate to master the skills behind this story.

More in Tools

Anthropic Launches Claude Design, Turning Text Prompts Into Slides, Prototypes and One-Pagers
Tools

Anthropic Launches Claude Design, Turning Text Prompts Into Slides, Prototypes and One-Pagers

Anthropic introduced Claude Design on April 17, 2026, a research preview that converts text descriptions into shareable visuals like prototypes, slides and one-pagers using Claude Opus 4.7.

15 hours ago2 min read
Google Brings AI Mode Side-by-Side With Web Pages in Chrome
Tools

Google Brings AI Mode Side-by-Side With Web Pages in Chrome

Google's Chrome desktop now keeps AI Mode open alongside web pages, lets users query across multiple tabs and PDFs at once, and surfaces image and Canvas tools through a new plus menu.

1 day ago2 min read
Canva AI 2.0 Launches as Agentic Design Platform With Proprietary Proteus, Lucid Origin, and I2V Models
Tools

Canva AI 2.0 Launches as Agentic Design Platform With Proprietary Proteus, Lucid Origin, and I2V Models

Canva unveiled Canva AI 2.0 on April 16 as a research preview, recasting its design tool as an agentic workspace powered by three proprietary models the company claims are up to 7x faster and 30x cheaper than frontier alternatives.

2 days ago2 min read