Codex from 0 to 10M Users: Building ChatGPT Work — Akshay Nathan, OpenAI
Codex usage has exploded, with MAU increasing over 10x from January 2026, reaching 10 million users combined with ChatGPT Work shortly after launch. OpenAI reorganized to integrate Codex and ChatGPT Work under a unified "Superapp," shifting focus from coding agents to knowledge work agents. Knowledge workers now account for roughly 20% of Codex's user base, growing three times faster than developers, indicating a significant shift in AI adoption beyond traditional coding. ChatGPT Work enables us
Analysis
TL;DR
- Codex usage has exploded, with MAU increasing over 10x from January 2026, reaching 10 million users combined with ChatGPT Work shortly after launch.
- OpenAI reorganized to integrate Codex and ChatGPT Work under a unified "Superapp," shifting focus from coding agents to knowledge work agents.
- Knowledge workers now account for roughly 20% of Codex's user base, growing three times faster than developers, indicating a significant shift in AI adoption beyond traditional coding.
- ChatGPT Work enables users to collaborate with agents across various tools and contexts, simplifying tasks that previously required manual operation of multiple applications.
- The shared agent harness behind Codex and ChatGPT Work allows for persistent environments, artifacts, Sites, plugins, memory, and sub-agents, enhancing the delegation of tasks to AI.
Why It Matters
This development is crucial for AI practitioners and researchers as it highlights the rapid expansion of AI tools beyond their initial coding applications into broader knowledge work. The integration of Codex and ChatGPT Work demonstrates a strategic move towards creating more versatile and accessible AI tools that can cater to a wider audience, including non-developers. This trend suggests a future where AI becomes an integral part of everyday productivity, transforming how tasks are managed and completed across various industries.
Technical Details
- Shared Agent Harness: Both Codex and ChatGPT Work utilize a common underlying agent framework, allowing for seamless integration and consistent user experience while maintaining distinct functionalities tailored to different user needs.
- Persistent Environments: The introduction of persistent environments, such as Sites and artifacts, enables users to maintain continuity and context across sessions, enhancing the efficiency and effectiveness of AI-assisted tasks.
- Memory and Context Gathering: Advanced memory capabilities allow ChatGPT Work to gather and retain context from various sources like code, Slack messages, documents, and local files, facilitating more informed and relevant assistance.
- Sub-Agents and Task Delegation: The use of sub-agents allows for complex tasks to be broken down into manageable parts, with each sub-agent handling specific aspects, thereby improving the overall performance and reliability of the system.
- Sandboxing and Security: Enhanced sandboxing mechanisms ensure that AI-generated actions are contained within secure environments, mitigating potential risks and ensuring user data integrity.
Industry Insight
- Shift Towards Generalist Skills: As AI tools become more accessible, there will be a growing demand for individuals who possess a broad skill set with specialized expertise, enabling them to leverage these tools effectively across various domains.
- Emphasis on Creativity and Judgment: With AI capable of generating content and performing routine tasks, the value of human creativity, critical thinking, and decision-making will increase, making these skills highly sought after in the workforce.
- Integration of AI into Daily Workflows: The success of tools like ChatGPT Work indicates a trend towards deeper integration of AI into daily workflows, potentially leading to increased productivity and innovation across industries. Organizations should consider how they can adopt and adapt these tools to enhance their operations.
Disclaimer: The above content is generated by AI and is for reference only.