OpenAI agents accelerate internal workflows

OpenAI's internal metrics prove that AI agents are no longer just speculative dev tools, but have become the core operational backbone of tech companies. The transition from transient chatbot queries to multi-hour autonomous tasks marks the dawn of the agentic enterprise.

• –**99.8% token share**: Codex eclipsing ChatGPT internally shows that developers and business teams alike prefer background delegation over interactive prompting.
• –**Beyond engineering**: Widespread adoption in Legal and Recruiting proves agentic frameworks have matured enough to handle complex, non-technical organizational tasks.
• –**Long-horizon delegation**: Over a quarter of surveyed user requests exceeded eight hours of human labor, indicating a major shift in how human-AI workloads are budgeted.
• –**Orchestration complexity**: Top users running over 60 hours of agent execution daily via parallel agents shows that managing agent swarms is the new programming paradigm.