explainx / blog
OpenAI launches Sites for creating interactive web apps and six role-specific plugins bundling 62 business tools and 110 skills—expanding Codex from 5 million developers to analysts, marketers, and business professionals growing 3x faster.
Jun 25, 2026
On June 25, 2026, Higgsfield CEO Alex Mashrabov introduced Supercomputer 2.0 — the company's first autonomous marketing agent, built on NVIDIA's Agent Toolkit and Nemotron subagents. Team and Enterprise plans, Inc. coverage, and a PSA Skincare case study with 29× views and 37× likes.
Jun 19, 2026
Show Codex your process once. It records it, learns your preferences, and creates a skill it can reuse—via computer use, browser use, or plugins. No prompts, no repeated instructions.
Jun 18, 2026
Unreal Engine 5.8 (June 17, 2026) connects LLM agents to the Editor via MCP. Grummz showed Claude and Codex beside the engine controlling Blueprints, PCG, and lighting—Epic's first-party Toolset plus a growing plugin ecosystem.
On June 2, 2026, OpenAI announced a major expansion of Codex—but this time, not for developers.
The company introduced Sites, a feature that lets teams create interactive, hosted websites and apps using natural language, and six role-specific plugins that bundle 62 business applications and 110 automated skills for knowledge workers.
The message is clear: Codex is no longer just a coding assistant. It's becoming a general-purpose AI platform for business professionals.
And the adoption numbers back it up. More than 5 million people now use Codex every week. Non-developers—analysts, marketers, operators, designers, researchers, investors, and bankers—make up about 20% of overall users and are growing more than 3x as fast as developers.
| Topic | Key Facts |
|---|---|
| Sites Feature | Create interactive, hosted web apps and dashboards from natural language; available in preview for Business/Enterprise. |
| Role-Specific Plugins | Six plugins bundle 62 apps (Snowflake, Figma, Salesforce) and 110 skills for analytics, sales, creative, design, finance. |
| Annotations | Select and refine specific parts of documents, spreadsheets, slides, and sites without starting over. |
| User Growth | 5 million weekly users; non-developers now 20% of base and growing 3x faster than developers. |
| Availability | Sites in preview for Business/Enterprise; plugins rolling out in supported regions via Codex plugin directory. |
| Partner Ecosystem | Vercel, Wix, Base44, Replit, Lovable, Figma, Webflow, Emergent working on Sites partner ecosystem. |
Sites is the headline feature. It lets Codex users create interactive, hosted websites and apps that can be shared via URL with anyone in their workspace.
This isn't just static content generation. Sites can include:
According to OpenAI's blog post, "Instead of adapting work to the limits of a single tool or file, teams can create sites that fit the work."
Sites are rolling out in preview for Business and Enterprise customers. Enterprise admins can enable Sites in admin settings.
OpenAI is launching role-specific plugins that bundle the relevant tools, instructions, and workflows for specific job functions. Each plugin aggregates multiple apps and skills, so users don't have to configure integrations manually.
The six initial plugins:
Helps analysts and business teams answer questions with data. Features:
Helps marketing and creative teams turn briefs into assets. Features:
Helps sales teams bring customer context into the work that moves deals forward. Features:
Built for turning early ideas into prototypes teams can review. Features:
Helps investors make sense of market and company information. Features:
Helps bankers turn research and diligence into client-ready materials. Features:
Together, these plugins include 62 popular apps and 110 skills.
Annotations extend the way developers already use Codex to refine code, Markdown, and websites—but now for documents, spreadsheets, and slides.
The workflow:
OpenAI explains: "Annotations make Codex more useful after the first draft, when the work needs judgment, feedback, and iteration."
Technically, when you highlight a segment—like a block of cells in a financial model—Codex isolates those exact data arrays and executes code strictly within that boundary.
OpenAI is making a strategic pivot from developer-first to enterprise-wide.
This mirrors the trajectory of tools like Notion, Airtable, and Figma—products that started technical but expanded to business users by solving adjacent problems.
OpenAI reports that internally, non-technical teams use Codex to:
Zapier: Teams use Codex to pull knowledge from tools like Slack, Google Docs, and Coda, then turn that context into postmortems, incident response plans, and feature tickets.
NVIDIA: Researchers are using Codex to speed up experiment workflows, from finding research ideas to writing scripts for machine learning infrastructure.
OpenAI is building toward an open ecosystem where partners can create and deploy their own plugins directly in Codex and ChatGPT.
Early Sites partners include:
This suggests Sites will evolve into a platform, not just a feature—similar to how Notion or Retool enable third-party integrations.
Claude Code (Anthropic's developer-focused CLI) has no equivalent to Sites and remains focused on terminal-based coding workflows. Anthropic's strategy emphasizes managed agents and financial services plugins, not horizontal expansion.
Microsoft 365 Copilot integrates deeply with Word, Excel, PowerPoint, and Teams but doesn't let you create custom hosted apps. Sites fills a gap between Copilot's integrations and Power Apps' complexity.
Replit Agent lets you build and deploy apps from natural language, but it's aimed at developers building software projects, not business users creating dashboards or scenario planners.
OpenAI's approach is horizontal: one AI tool that adapts to multiple roles through plugins and lets users create custom tools through Sites.
Based on OpenAI's examples and early customer feedback, here are use cases:
Annotations solve a major pain point: the "all or nothing" problem of AI-generated content.
If Codex generated a 10-slide deck and slide 7 needed a chart update, you had two options:
This makes iterative refinement practical.
According to OpenAI's blog post, Codex maps a document's underlying data schema. When you highlight a segment—like a block of cells in a spreadsheet—Codex isolates those exact data arrays and executes code strictly within that boundary.
This ensures:
OpenAI announced more role-specific plugins are coming soon:
The roadmap suggests OpenAI is targeting high-value knowledge work where automation can save hours per task.
OpenAI states: "This is just the start: we're building toward an open ecosystem where partners can create and deploy their own plugins directly in Codex and ChatGPT."
This mirrors Slack's app directory, Notion's integrations, and Zapier's connector ecosystem—playbooks that turned point products into platforms.
OpenAI notes: "Teams can also adapt [plugins] to their workflows or build and share custom plugins for their own systems and processes."
This suggests an API or SDK for creating custom plugins, though OpenAI hasn't detailed the implementation yet.
This expansion has major implications for OpenAI's revenue strategy.
OpenAI's business model has relied on:
Adding enterprise seats for non-developers creates a third revenue stream: per-seat SaaS pricing similar to Microsoft 365, Salesforce, or Slack.
If 20% of Codex's 5 million weekly users are non-developers, that's 1 million knowledge workers. If even a fraction convert to paid Business/Enterprise seats at $30-60/month, that's $30M-$60M in annual recurring revenue from a single user segment.
Sites directly competes with:
Role-specific plugins compete with:
If Codex can bundle these capabilities into one AI interface, it could unbundle specialized SaaS tools—a pattern we've seen before with AI-native startups.
For developers building with AI, this announcement offers lessons:
OpenAI is betting that one AI tool that adapts to multiple roles is better than separate tools for each role. This requires:
Sites proves that natural language can replace traditional app builders for certain use cases. This aligns with forward-deployed engineering trends where AI-assisted workflows replace manual configuration.
The ability to refine AI outputs in place is a major improvement over "regenerate everything." If you're building AI products, consider:
OpenAI is positioning plugins as the interface layer between AI and third-party tools. This mirrors:
If you're building AI tools, consider plugin-first architecture over custom integrations.
As we've covered in our analysis of AI token costs, enterprise AI deployments face two major challenges:
Microsoft banned Claude Code due to token costs spiraling out of control. If Codex Sites generate complex web apps with large codebases, token usage could balloon.
Sites could lead to sprawl: teams creating dozens of one-off tools that become hard to maintain, audit, or secure.
Enterprise IT teams will need:
If you're building in the AI space, this announcement is both opportunity and threat.
If your startup is:
...you now compete directly with OpenAI.
If OpenAI opens up plugin development, there's a land grab for:
The first movers in OpenAI's plugin directory could capture distribution similar to early Shopify apps or Slack bots.
OpenAI's Codex expansion is bigger than it looks.
On the surface, it's a feature release: Sites, plugins, annotations. But strategically, it's a pivot from developer tool to enterprise platform.
The bet:
If it works, OpenAI doesn't just power ChatGPT—it becomes the operating system for knowledge work.
For developers, the lesson is clear: the AI layer is moving up the stack. Tools that required configuration, code, or integrations are being replaced by natural language interfaces that "just work."
The winners will be those who build on top of this shift, not against it.
AI tools and enterprise features evolve rapidly. Features and pricing are based on public announcements through June 2026. For current Codex availability, visit OpenAI Codex.
