When Codex launched, the assumption was that it would live and die as a tool for engineers. The numbers are correcting that assumption fast. More than 5 million people now use Codex every week, and non-developers — analysts, designers, investors, bankers, marketers, and operators — account for about 20% of that user base while growing more than 3x faster than developers. On June 2, 2026, OpenAI shipped six role-specific plugins, a Sites feature that lets Codex build and host interactive web apps, and an annotations system that makes post-draft editing far more precise. That is a meaningful product direction change, not just a feature add.

Six Plugins That Map to Real Job Functions

The core idea is straightforward: instead of asking every knowledge worker to learn how to prompt for their specific use case, OpenAI built plugins that bundle the right apps, skills, and workflows for each role out of the box. Today’s launch covers six verticals, and the integrations list is specific enough to take seriously.

• Data analytics: Lets analysts and business teams query product and business data, diagnose why key metrics changed, and build dashboards. Integrations include Snowflake, Databricks Genie, Hex, and Tableau, with more arriving soon.
• Creative production: Helps marketing and creative teams turn a brief into reviewable assets. Covers campaign boards, display ad variations, and product image sets via Figma, Canva, Shutterstock, Picsart, and Fal.
• Sales: Surfaces account signals, preps for customer meetings, completes follow-ups, updates CRM records, and flags deals at risk. Tools include Salesforce, HubSpot, Slack, Outreach, Clay, Rox, and Actively.
• Product design: Focused on rapid prototyping from early ideas to reviewable mockups. Teams can audit user flows, prototype from live URLs, and carry work forward in Figma and Canva.
• Public equity investing: Helps investors review earnings, compare companies, and track signals using data from Moody’s, Daloopa, Datasite, FactSet, LSEG, S&P, PitchBook, and Hebbia.
• Investment banking: Supports pitch materials, comparable company analysis, and turning diligence into client-ready recommendations using trusted financial data sources.

Across all six roles combined, the plugins connect to 62 popular apps and bundle 110 distinct skills. Teams can customize any plugin or build and share their own for internal systems. OpenAI confirmed that Corporate Finance, Private Equity Investing, Marketing Strategy, Strategy Consulting, and Legal plugins are coming in a subsequent wave.

What Real Teams Are Already Doing with Codex

The launch post gives three concrete examples of how non-developer teams are using Codex in production today. At Zapier, teams pull knowledge from Slack, Google Docs, and Coda, then turn that context into postmortems, incident response plans, and feature tickets. At NVIDIA, researchers use Codex to accelerate experiment workflows, from identifying research directions to writing scripts for machine learning infrastructure. Inside OpenAI itself, non-technical teams use it to build internal apps, prepare executive materials, create dashboards, and turn creative briefs into outputs that reflect brand and design constraints.

These are not hypothetical demos. They are production workflows at companies with access to every other productivity tool on the market. The fact that these teams chose Codex anyway says something about where the tool is actually delivering value.

“Non-developers — including analysts, marketers, operators, designers, researchers, investors, and bankers — make up about 20% of overall Codex users and are growing more than 3x as fast as developers.”

— OpenAI, Codex product announcement, June 2, 2026

Sites: Codex as a Web Builder

The most genuinely new capability in this release is Sites, currently in preview for Business and Enterprise customers. Codex can take ideas, analysis, or plans and generate interactive, hosted web apps that teams share via a URL inside their workspace. The examples OpenAI gives are specific: a customer review page with product updates, open questions, usage trends, and next steps per account; a scenario planner built from a financial model so leaders can compare assumptions instead of reading through spreadsheet tabs; a launch hub tracking messaging, milestones, owners, and decisions in one living place.

OpenAI is building a partner ecosystem for Sites that already includes Vercel, Wix, Base44, Replit, Lovable, Figma, Webflow, and Emergent. That list is notable. Vercel and Replit are developer-first platforms, but Wix and Figma are mainstream design and website tools. The intent is clearly that Sites should be approachable for non-developers who want to ship something interactive without writing code.

For context, this is a meaningful parallel to what Claude Opus 4.8 introduced with Dynamic Workflows on the Anthropic side: both companies are racing to make their flagship products useful for people who are not writing code. The competitive pressure is visible in the product roadmaps on both sides.

Annotations: Editing After the First Draft

Annotations extend an existing developer-focused feature to all content types. Developers have been able to point to a specific part of a file or website Codex generated and ask for a targeted change. That same workflow now applies to documents, spreadsheets, and slides. You can select a navigation bar and ask for a font update, highlight a claim in an investment thesis and ask where it came from, or mark a chart in a slide and request a clearer label.

The practical value is making iteration faster without burning the whole document. First-draft quality matters less when refinement is this precise. This is the kind of feature that sounds minor but changes day-to-day behavior significantly.

What This Means for Developers Building with Codex

The plugin architecture is open. Teams can build and share custom plugins for their own systems and deploy them directly across Codex and ChatGPT. For anyone tracking how Model Context Protocol is reshaping how agents connect to tools, the direction here is similar: give the model structured access to systems the user already relies on, and let it do more useful work without additional configuration from the end user.

The business implication is real. If Codex becomes the layer that analysts, bankers, and designers use to interact with Snowflake, Figma, or Salesforce, OpenAI is not competing with coding tools anymore. It is competing with every enterprise productivity platform. The Sites feature makes this even clearer: by letting teams generate and host lightweight web apps inside Codex, OpenAI is building a surface area that used to require dedicated no-code tools or a developer on hand.

For model comparison context, the Claude Opus 4.8 vs GPT-5.5 coding comparison put these two models on roughly similar footing for developer tasks. The Codex plugin expansion is OpenAI’s bet that the next wave of AI adoption comes from roles that never wrote a line of code, and that whoever owns that surface wins the enterprise.

The Bottom Line

OpenAI Codex reaching 5 million weekly users while launching six role-specific plugins, a built-in web app builder, and a precision editing system in a single update is a clear statement about where the product is heading. This is no longer a developer tool that occasionally gets used by other teams. It is an enterprise productivity layer that happens to be built on the same model engineers use to write code. For developers building on top of OpenAI’s platform, the custom plugin ecosystem is the most interesting near-term opportunity: the ability to deploy plugins across Codex and ChatGPT is a distribution channel that did not exist six months ago, and the 5 million active users are already there waiting.