OpenAI launches Codex app to manage multiple AI agents across software projects

AI Agents News

OpenAI launches Codex app to manage multiple AI agents across software projects
AutomationCodex AppCoding Tools
  • 📰 IntEngineering
  • ⏱ Reading Time:
  • 170 sec. here
  • 11 min. at publisher
  • 📊 Quality Score:
  • News: 95%
  • Publisher: 63%

OpenAI launches the Codex desktop app to help developers manage multiple AI agents working on software projects.

OpenAI has launched a new Codex desktop app aimed at helping developers manage multiple AI agents working in parallel across long-running software projects.The macOS app acts as a command center where developers can delegate tasks to different agents, supervise progress across projects, and review code changes without losing context.

OpenAI says the tool reflects a shift in how software is built, moving from single-agent assistance to coordinated teams of AI agents handling end-to-end development.Codex was first introduced in April 2025, but OpenAI says agent capabilities have expanded quickly. Models can now handle complex tasks that span hours, days, or even weeks, creating the need for tools designed specifically for oversight, collaboration, and control at scale.OpenAI CEO Sam Altman said the underlying model powering Codex represents a bigger leap than performance benchmarks may suggest. “I love building with this model; it feels like more of a step forward than the benchmarks suggest,” Altman wrote on X.The Codex app allows agents to run in separate threads organized by project. Developers can inspect diffs, comment directly on changes, or open work in their editor for manual edits. Built-in worktree support lets multiple agents work on the same repository simultaneously, each operating on an isolated copy of the code to avoid conflicts.From coding to controlCodex is also expanding beyond code generation through customizable skills. These allow agents to connect to external tools, follow predefined workflows, and complete tasks that involve research, writing, deployment, or image generation.GPT-5.3-Codex is here!*Best coding performance .*Mid-task steerability and live updates during tasks.*Faster! Less than half the tokens of 5.2-Codex for same tasks, and >25% faster per token!*Good computer use.— Sam Altman February 5, 2026Using skills, Codex can pull design assets from Figma, manage bug tracking and releases in Linear, deploy applications to cloud platforms such as Vercel and Cloudflare, and generate images using OpenAI’s image model.Developers can explicitly instruct Codex to use a skill or allow it to decide automatically based on the task.OpenAI demonstrated Codex’s autonomy by having it build a full 3D voxel racing game from a single prompt. The agent handled design, development, testing, and iteration by repeatedly playing the game itself to identify missing features and bugs.Altman said internal use of Codex has already accelerated development at OpenAI.“It was amazing to watch how much faster we were able to ship 5.3-Codex by using 5.3-Codex, and for sure this is a sign of things to come,” he wrote.The app also introduces Automations, which allow Codex to run scheduled background tasks such as issue triage, bug detection, and release summaries. Completed tasks are sent to a review queue so developers can step in only when needed.Security and scale focusSecurity is built into the Codex stack by default. Agents are sandboxed and limited to specific folders or branches, with permission required for elevated actions such as network access. OpenAI says this is its first model to score highly on internal cybersecurity readiness.“This is our first model that hits ‘high’ for cybersecurity on our preparedness framework,” Altman wrote, adding that OpenAI is piloting a Trusted Access framework and committing $10 million in API credits to support cyber defense efforts.The Codex app is available now on macOS for ChatGPT Plus, Pro, Business, Enterprise, and Edu users, with temporary access also extended to Free and Go users. OpenAI says Codex usage has doubled since mid-December, with more than a million developers using it in the past month.

We have summarized this news so that you can read it quickly. If you are interested in the news, you can read the full text here. Read more:

IntEngineering /  🏆 287. in US

Automation Codex App Coding Tools Enterprise AI Multi-Agent Systems Openai Software Development

 

United States Latest News, United States Headlines

Similar News:You can also read news stories similar to this one that we have collected from other news sources.

Super Bowl commercial sees Anthropic mock OpenAI for bringing ads to ChatGPTSuper Bowl commercial sees Anthropic mock OpenAI for bringing ads to ChatGPTAnthropic's 2026 Super Bowl LX ad
Read more »

OpenAI says it will use Google's cloud for ChatGPTOpenAI says it will use Google's cloud for ChatGPTHaving faced supply shortages, OpenAI will now be drawing on computing resources from CoreWeave, Google, Microsoft and Oracle.
Read more »

OpenAI Frontier is a single platform to control your AI agentsOpenAI Frontier is a single platform to control your AI agentsOpenAI launches its Frontier platform to manage AI agents for enterprises.
Read more »

OpenAI exec is still 'bullish' about the future of softwareOpenAI exec is still 'bullish' about the future of softwareBusiness Insider tells the global tech, finance, stock market, media, economy, lifestyle, real estate, AI and innovative stories you want to know.
Read more »

California AI rivals Anthropic and OpenAI to battle with Super Bowl adsCalifornia AI rivals Anthropic and OpenAI to battle with Super Bowl adsOpenAI and Anthropic, the startups behind ChatGPT and Claude, are gearing up for a major showdown
Read more »

AI Enters Era of Self-Improvement: OpenAI Releases Coding Model That Assisted in Its Own CreationAI Enters Era of Self-Improvement: OpenAI Releases Coding Model That Assisted in Its Own CreationOpenAI unveils GPT-5.3-Codex, a cutting-edge coding model that played a key role in its own development. The AI model's ability to debug, manage deployment, and analyze its training data is accelerating AI development, blurring the lines between engineers and designers and raising questions about AI safety and human roles.
Read more »



Render Time: 2026-04-01 13:34:07