OpenAI Expands Codex Into Enterprise Work Platform
OpenAI expanded Codex with hosted web apps, in-place editing, and role-specific plugins that connect 62 business apps. The company said non-developers now make up about 20% of Codex’s 5 million weekly users and are adopting it three times faster than engineers.
OpenAI on Tuesday expanded Codex from a coding agent into a broader enterprise work platform, adding tools aimed at workers outside software engineering as well as developers.
The update brings three main features: Sites, which lets users create and share hosted interactive web applications; Annotations, an in-place editing tool; and six role-specific plugins that connect Codex to 62 business applications and include 110 automated skills. OpenAI said the changes are meant to make Codex useful for more kinds of office work, not just programming.
⚡ New to this?
This matters because Codex is moving beyond software development and into ordinary office tasks like reporting, planning, and connecting business apps. A plugin is a connector that lets one system talk to another, and an orchestration layer is the tool that coordinates work across those systems.
OpenAI is saying that people who do not write code are now using its product faster than engineers. That suggests AI tools are starting to change how non-technical teams build and use software inside companies.
🦞 OpenClaw angle
If you run self-hosted AI automation, treat this as a signal to build for non-developers, not just engineers. Focus on natural-language workflows that can turn internal data into shareable dashboards or simple web apps, because that is where adoption is growing fastest.
Also, prioritize connectors and permissions. If your agent can act across business systems, make sure each action is logged, role-scoped, and easy to audit so operations, finance, or marketing teams can use it safely.
Finally, design your agent layer to sit above existing tools rather than replacing them outright. The winning pattern here is orchestration: one interface that can read from multiple systems, produce an output, and hand control back to the human when needed.
Sites is the most visible addition. In preview for business and enterprise customers, it allows Codex to turn prompts into interactive web apps that can be shared through secure workspace URLs. According to OpenAI’s description, that could let a financial analyst turn a spreadsheet into a live dashboard, scenario planner, or interactive model that co-workers can use without downloading a file.
The company is also pushing Codex deeper into enterprise software workflows through its plugin system. The six role-specific plugins connect to business products including Snowflake, Figma, and Salesforce, and OpenAI said they come with 110 automated skills built in.
That positions Codex as a layer above existing SaaS tools rather than a replacement for them. A marketing manager, for example, could use Codex’s natural-language interface to work across multiple connected services instead of switching between each app separately.
OpenAI’s announcement also included usage data that shows how quickly the product is spreading beyond engineering teams. The company said non-developers now account for about 20% of Codex’s 5 million weekly users and are adopting the platform three times faster than engineers.
Those non-developers include financial analysts, marketers, operations staff, and researchers, according to OpenAI. The company’s data suggests that so-called vibe coding - building applications through natural-language prompts rather than hand-written code - is moving from an experiment to a measurable part of enterprise software use.
The broader strategy is clear from the product direction. OpenAI is not only trying to sell Codex as a coding assistant, but as a place where employees can create, edit, and connect work tools across departments.
That puts Codex into competition with parts of the workflow stack occupied by business intelligence tools and internal app builders. It also reflects a wider industry push from companies such as Salesforce and Microsoft, which are building agent layers that sit on top of existing software and coordinate work across multiple systems.
Whether enterprises adopt Codex at scale will depend on how well the preview version handles reliability, security, and auditability. For now, OpenAI is making a clear bet that the same AI system used to write code can also become an interface for everyday business work.
