To say there are whispers in the tech industry of OpenAI’s ambition to build a so-called super app would be wrong. No, OpenAI’s plans for a unified super app are very much public, and as Denise Dresser, Chief Revenue Officer, OpenAI, says, will be “one place where employees can work with AI agents throughout the day to complete tasks and take action across the tools they already use,” are very much in motion.
In fact, as I reported earlier this month, the company has earmarked a portion of its recent $122 billion funding round to work toward realizing its super app, combining the powers of ChatGPT, Codex, browsing, and OpenAI’s wider agentic capabilities. Now, a series of major updates to Codex, OpenAI’s software engineering agent — or AI coding assistant — is making the dream of the super app closer to becoming a reality.
New Features
“Our mission is to ensure that AGI benefits all of humanity. That includes narrowing the gap between what people can imagine and what they can build. This release brings Codex closer to the tools, workflows, and decisions involved in building software, with much more to come soon,” reads an OpenAI blog post.
Ultimately, OpenAI has extended the capabilities of Codex to achieve numerous tasks beyond coding. A new background computer use feature enables Codex to access any app on a user’s computer in the same way a human operator does, “by seeing, clicking, and typing with its own cursor.”
This functionality isn’t limited to a single agent; multiple agents can work in parallel without interrupting any workflows undertaken by others, including the primary operator — the human. Codex can now work natively with the web, featuring an in-app browser that allows users to comment directly on webpages to provide precise instructions to the agent.
Another new feature is Codex’s ability to generate and iterate on images using gpt-image-1.5. And OpenAI is releasing over 90 additional plugins — including GitLab Issues, Microsoft Suite, and Neon by Databricks — combining skills, app integrations, and MCP servers, providing Codex with more ways to construct context and take actions.
OpenAI has introduced a series of dedicated support functions for GitHub review comments, operating multiple terminal tabs, connecting to remote devboxes over SSH in alpha, a new summary pane for tracking agent plans, sources, and artifacts, and the ability to preview files in the sidebar.
Codex users also now benefit from expanded automations, enabling them to reuse existing conversation threads with context preserved and to leverage Codex’s ability to schedule and carry out work for itself, automatically waking up to do it.
Codex memory, currently in preview, enables the agent to remember context from previous experiences, such as individual preferences, corrections, and information accumulated over time. It can also suggest the best place to start working again by tapping into the context alongside memory, and plugins to help users to continue from where they left off.
Closing Thoughts
Not long ago, the gold standard in the super app space was WeChat, China’s answer to universal convenience with the ability to enable payments, social media, and messaging. However, in the AI Era, the expectations have shifted exponentially. Now, what OpenAI is working toward is redefining what a super app is by unleashing agents that not only answer and act, but make decisions and learn from the user at the helm.
