From Coding Agent to Codex Enterprise Platform
Codex is OpenAI’s agentic AI platform that began as a coding assistant and is now expanding into a full enterprise system for software, research, analysis, and other knowledge work tasks across professional roles. Codex has reached 5 million weekly active users, a sixfold increase since its desktop app debuted in February, and its user base is changing fast. Around 20 percent of users are now knowledge workers, adopting Codex more than three times as fast as developers. Personal users already exceed 5 percent of the total and are growing more than four times as fast as software engineers. The task mix has moved beyond pure code: each week, most knowledge-worker users create reports, memos, PDFs, and spreadsheets, and many run parallel projects, such as inspecting datasets, drafting scripts, and assembling market research in separate threads. Codex is no longer a niche developer tool; it is becoming an AI knowledge work platform.
AI Knowledge Work Tools: Reports, Research, and Spreadsheets
Codex now sits in the middle of day-to-day knowledge work, not only in software teams but also in operations, marketing, product, and research roles. According to Help Net Security, roughly 40 percent of labor works primarily with information, and Codex is increasingly taking over their repetitive tasks. Seventy-two percent of knowledge workers using Codex each week produce artifacts such as reports, contracts, images, videos, PDFs, and spreadsheets, blurring the line between engineering and business work. Data analysis is the fastest-growing category among these users, expanding 110 percent week over week, followed by research and knowledge artifact creation. Within this, work with PDFs and spreadsheets is growing more than 50 percent. Market and competitive research, data labeling, drafting messages, hiring workflows, and understanding contracts and regulations are all gaining share, turning Codex into a daily AI knowledge work tool rather than a side assistant.
OpenAI Codex Plugins Target Banking, Analytics, and Creative Jobs
OpenAI is pushing beyond generic chat by releasing OpenAI Codex plugins built for specific jobs. Six role-focused plugins now target data analytics, creative production, sales, product design, equity investing, and investment banking, each bundling integrations, workflows, and domain instructions so non-technical professionals can use Codex out of the box. Collectively, the initial plugin set connects to 62 enterprise applications and 110 specialized skills, from analytics tools such as Snowflake, Databricks Genie, Hex, and Tableau to creative and productivity platforms. In finance, new public-equity-investing and investment-banking plugins give professionals AI that can analyze earnings, compare companies, track signals, and evaluate whether investment theses are strengthening or deteriorating using institutional-grade data from Moody’s, FactSet, LSEG, PitchBook, Daloopa, Datasite, and S&P. These AI knowledge work tools show a shift toward vertical, profession-aware systems that understand the structure of real jobs in banking, consulting, and corporate advisory.
Sites, Annotations, and the New AI Workspace for Business Professionals
To serve business professionals beyond engineering, OpenAI has added new Codex features that turn the agent into a shared workspace. Sites lets users publish work as interactive websites inside Codex, shared via URL within a workspace. Teams can build custom dashboards, launch hubs, scenario planners, or internal tools that sit on top of financial models and live data, instead of forcing work into static documents. Partners such as Wix, Base44, Replit, Lovable, Figma, and Emergent extend what these sites can do, from web front-ends to design prototypes. Annotations gives more precise control by letting users point Codex to specific parts of documents for targeted changes or analysis, which is especially useful for contracts, policies, and long reports. Together with job-specific plugins, these capabilities turn Codex into an AI workspace where sales, marketing, operations, and product teams can create, refine, and share tools, not just text responses.
Enterprise AI Deployment: AWS, Bedrock, and Desktop Control
Codex’s shift into the Codex enterprise platform is backed by a push to meet security, compliance, and governance needs. OpenAI has brought Codex and its frontier models, including GPT-5.5, into Amazon Web Services through Amazon Bedrock, letting enterprises deploy AI through cloud infrastructure, identity, billing, and governance systems they already trust. This reduces the friction of separate security reviews and procurement cycles and fits Codex into existing enterprise AI deployment strategies. For software and IT teams, Codex inside Bedrock supports AI-assisted coding, code review, debugging, and modernization as part of broader delivery pipelines. On the endpoint side, Codex Computer Use now supports Windows, so developers can trigger sandboxed desktop sessions from phones or other devices to control local tasks safely. Combined with new job-specific plugins and Sites, these integrations position Codex as an AI for business professionals operating across cloud, desktop, and internal tools.
