OpenAI is turning Codex from a software-development assistant into a broader work system. In a June 2, 2026 product post, the company said more than 5 million people now use Codex every week. It also said non-developers make up about 20% of overall Codex users and are growing more than three times as fast as developers.
The product changes match that user mix. OpenAI is adding role-specific plugins, Sites, and annotations so teams can connect Codex to business tools, create shareable work surfaces, and refine outputs in place. The headline is not that Codex can write more code. It is that OpenAI wants Codex to become a workbench for analysts, marketers, operators, designers, researchers, investors, bankers, and sales teams.
The user mix is the signal
Codex started as a coding product, but OpenAI’s usage numbers point to a different ambition. If non-developers are already about a fifth of users, then Codex is no longer only competing for developer workflows. It is competing with spreadsheets, slide decks, dashboards, project docs, design tools, and internal app builders.
OpenAI’s examples are deliberately business-shaped. The company says non-technical teams inside OpenAI use Codex to build internal apps, prepare executive materials, create dashboards, and turn creative briefs into work that follows brand and design constraints. It also cites Zapier using Codex to pull context from tools like Slack, Google Docs, and Coda, then turn that context into postmortems, incident response plans, and feature tickets.
That is a different product promise from “ask an AI to draft something.” Codex is being positioned as a system that can gather context, take action across tools, generate an artifact, and keep refining it.
Plugins make Codex more packaged
The new role-specific plugins are the clearest packaging move. OpenAI says the six plugins bundle relevant apps, skills, instructions, and workflows. Together, they include 62 popular apps and 110 skills.
The first set covers data analytics, creative production, sales, product design, public equity investing, and investment banking. That list is not random. These are functions where work often starts with scattered context and ends in an artifact: a dashboard, campaign board, deal plan, prototype, investment memo, or pitch material.
The practical test is whether plugins reduce setup work enough for business teams to use Codex without a technical operator nearby. If a sales team still needs someone to wire every connector and instruction manually, adoption will stay narrow. If the plugin gets most of the workflow ready out of the box, Codex becomes easier to deploy as a team product.
Sites turn outputs into shared surfaces
Sites are the other major shift. OpenAI says Codex can now create and share interactive hosted websites and apps in preview for business and enterprise customers. The examples include dashboards, planners, review workspaces, project boards, galleries, and lightweight tools that can be shared inside a workspace by URL.
This is where Codex starts to look less like a chat product and more like an internal software surface. A finance team can ask for a scenario planner. A launch team can ask for a live hub with messaging, milestones, owners, and decisions. A customer team can ask for a review page with product updates, open questions, usage trends, and next steps.
The risk is governance. Shareable generated sites are useful only if admins can control access, data permissions, and what underlying apps Codex can touch. OpenAI says Business and Enterprise admins can control underlying app permissions in workspace settings, and Enterprise admins can enable Sites in admin settings.
Annotations are the workflow repair layer
Annotations extend a behavior developers already know: point to a specific part of the output and ask Codex to change it. OpenAI says that now applies beyond code, Markdown files, and websites to documents, spreadsheets, and slides.
That may sound smaller than plugins or Sites, but it matters for real work. First drafts are rarely the bottleneck. The bottleneck is making the right edit without disturbing everything else. If annotations let a user select a chart label, a claim in an investment thesis, or a navigation bar in a generated site and revise only that piece, Codex becomes more useful after the first response.
