This week marked a turning point for OpenAI’s Codex platform โ it’s no longer just a coding assistant. The repositioning of Codex as a general-purpose daily work agent, combined with a new Chrome extension and the milestone of 4 million weekly active users, signals that OpenAI sees Codex as the primary interface for AI-augmented knowledge work. For enterprise teams evaluating agent platforms, the feature velocity alone demands attention.
Codex Becomes a Daily Work Agent
On May 1, OpenAI officially repositioned Codex beyond its coding roots. ChatGPT users can now connect Codex to Slack, Google Drive, email, and calendar for automating routine business tasks โ not just software development. CEO Sam Altman posted “Big upgrade for codex today! Try it for non-coding computer work,” and president Greg Brockman stated plainly that “Codex is for everyone, for any task done with a computer.”
Why this matters for enterprise teams: This changes the procurement conversation. If Codex can handle both developer workflows and knowledge-worker automation, organizations can standardize on a single agent platform instead of buying separate tools for engineering and business operations. The OpenAI Academy webinar on “Codex for everyday work” specifically targets non-developer use cases โ a signal that enterprise sales motion is expanding beyond tech audiences.
Codex Chrome Extension Launches
On May 7, OpenAI released Codex for Chrome, a browser extension enabling Codex to work directly in Chromium-based browsers on Mac and PC. The extension can test web applications, gather context across multiple tabs, access Chrome DevTools, and automate UI testing using natural language. Users control per-site access permissions, and Codex runs background tasks without taking over the active browser session.
Why this matters: The browser is where most knowledge work happens. A Codex agent that can read your tabs, interact with web apps, and run background tasks opens up automation scenarios that previously required brittle scripting or dedicated RPA tools. For consulting engagements, Big Hat Group’s AI automation practice helps enterprises identify and implement these automation opportunities. This makes Codex a viable candidate for testing automation, data extraction workflows, and cross-platform orchestration.
4 Million Weekly Active Users โ 8ร Growth Since January
On May 7, OpenAI announced that Codex has surpassed 4 million weekly active users โ an 8ร increase since the beginning of 2026. The platform added 1 million users in the last two weeks alone, reflecting accelerating enterprise and developer adoption.
Why this matters: These are the first public user metrics for Codex, and the growth curve is remarkable. For enterprise buyers, 4 million weekly users means Codex has crossed the chasm from early adopter tool to mainstream platform. The ecosystem of skills, extensions, and community resources will only accelerate from here.
Zero-Seat-Fee Enterprise Pricing
OpenAI launched Codex-only pay-as-you-go seats for ChatGPT Business and Enterprise with zero seat fees, billed purely per usage. This decouples Codex adoption from full ChatGPT seat licensing.
Why this matters for organizations: If you’ve been blocked by per-seat minimum commitments or don’t want to roll out ChatGPT to an entire department just to pilot Codex, this removes the friction. You can buy Codex capacity on its own, experiment with a small team, and scale based on observed ROI โ exactly the procurement pattern enterprise architects prefer.
Codex Security โ Research Preview
OpenAI introduced Codex Security, an AI application security agent that analyzes project context to detect, validate, and patch complex vulnerabilities. It’s available as a research preview.
Why this matters: For enterprises with security compliance requirements, this is a significant addition. An agent that can identify vulnerabilities in your codebase and propose patches within the same workspace streamlines the DevSecOps pipeline. It’s early-stage, but the direction is clear: OpenAI is building security directly into the agent platform rather than requiring third-party tooling.
gpt-image-2 Now in Codex
OpenAI released gpt-image-2 in both the API and Codex, purpose-built for production image generation workflows. Developers can generate and edit images directly through Codex sessions.
Simplex Case Study: AI-Driven Development at Scale
Simplex published a case study on rethinking software development with Codex, using ChatGPT Enterprise and Codex to validate AI-driven development and scale engineering productivity. For teams looking to build a business case for Codex adoption, this provides concrete metrics and implementation patterns.
AWS Bedrock Now Offers Codex
Amazon announced that Codex is available through Bedrock via the Codex CLI, desktop app, and VS Code extension โ building on last week’s multi-cloud partnership announcement. Enterprise customers can deploy Codex alongside OpenAI models within the AWS ecosystem with managed agent infrastructure.
For a deeper comparison of Codex across cloud providers, see our Codex on AWS Bedrock vs. Azure OpenAI decision framework.
Codex CLI: /goal Goes Viral, v0.129.0 Ships
The Codex CLI continued its rapid release cadence with two noteworthy updates.
/goal โ Autonomous Coding Loops (v0.128.0 Buzz Continues)
The /goal autonomous workflow feature, first shipped in v0.128.0, continued generating significant community attention through early May. It enables persistent autonomous coding loops โ users report sessions running 15โ18+ hours without intervention. Enabled via features.goals = true in ~/.codex/config.toml, currently available in CLI only.
Codex CLI 0.129.0 Released
Codex CLI 0.129.0 shipped in May with:
- Modal Vim editing support in the composer (
/vimcommand, custom keymaps) - Redesigned workflow resume/fork picker
- Raw scrollback mode
- Improved
/diffwith workspace awareness - Theme-aware status lines with optional PR/branch summaries
- Enhanced plugin management โ workspace sharing, granular access controls, remote bundle sync, admin plugin disable
- Hooks extensibility improvements
GPT-5.5 Powers Codex on NVIDIA Infrastructure
NVIDIA published a blog detailing how GPT-5.5 powers Codex on NVIDIA infrastructure, highlighting enterprise deployment patterns and NVIDIA’s internal use of Codex for AI agent workflows.
Community: Awesome Codex Skills Catalog
ComposioHQ launched awesome-codex-skills, a curated community catalog of practical Codex skills for automating workflows across the CLI and API. This provides a structured way to discover and share reusable skill definitions beyond simple prompt templates.
Claude Code to Codex Migration Guide
A comprehensive migration guide was published for teams moving from Anthropic’s Claude Code to OpenAI’s Codex, covering AGENTS.md conversion, skills, hooks, profiles, MCP, and verified CLI notes.
Notable GitHub Issues
- Memory leak (75GB+) โ Issue #20740 reports Codex memory growing to ~85GB during prolonged usage, requiring force-stop
- Windows sandbox profile leak โ Issue #21455 documents temporary Windows profile leaks in sandbox mode
- Hooks not running after resume โ Community reports of hooks failing to execute after session resumption
What to Watch
- Codex for non-developers: The repositioning as a daily work agent opens new enterprise sales motions. Watch for integration depth with Microsoft 365, Google Workspace, and Slack.
- Chrome extension ecosystem: If the extension gains traction, expect third-party Chrome extensions that extend Codex’s browser capabilities for domain-specific workflows.
- Zero-seat-fee pricing adoption: This removes a major procurement blocker. Organizations that passed on Codex due to seat commitments should re-evaluate.
- Codex CLI stability: The memory leak and sandbox issues are concerning for production use. Watch for hotfixes in the 0.129.x release line.
That’s Codex Weekly for May 8. Between the shift to daily-work agent, the Chrome extension launch, 4 million users, and continuous CLI velocity, OpenAI is executing aggressively on its agent platform vision. Explore how Big Hat Group can help your team evaluate and deploy Codex โ from strategy and architecture to hands-on implementation. We specialize in AI agent systems on Azure, AWS, and Windows 365.
โ Kevin Kaminski, Principal Architect at Big Hat Group. We help enterprises design and deploy AI agent systems on Azure, AWS, and Windows 365.
Note: Three of four planned research briefs (ChatGPT/API, Agents/Enterprise, Company Ecosystem) were unavailable this week. This edition is compiled from the Core & CLI research file only.