OpenAI released new research from its B2B Signals program detailing how frontier enterprises are pulling ahead in artificial intelligence adoption. According to the data, firms at the 95th percentile of usage now consume 3.5 times as much intelligence per worker as typical firms, an increase from 2 times a year ago. The research indicates that message volume accounts for only 36 percent of the gap, with the majority of the advantage coming from richer, more complex AI utilization. Advanced and agentic tools show the widest adoption margins, with frontier companies sending 16 times as many Codex messages per worker as typical organizations. The findings also highlight real-world implementations, such as Cisco utilizing Codex to reduce build times and save engineering hours, and Travelers Insurance deploying an AI Claim Assistant to handle policy questions and claims processing. OpenAI's newly introduced B2B Signals platform aims to provide recurring, de-identified measurements of how AI is diffusing across businesses, tracking tools, tasks, and cross-industry function use cases. The report suggests that leading organizations are transitioning from simple chat-based assistance to delegating multi-step tasks to AI agents, redesigning workflows to build durable competitive advantages.
How frontier firms are pulling ahead
OpenAI has introduced B2B Signals, a new research initiative tracking how frontier enterprises deepen AI adoption and scale agentic workflows. Data shows top-tier firms now use 3.5 times as much intelligence per worker compared to typical organizations. The advantage is driven by complex tool usage, such as Codex, rather than simple message volume.
Why it matters
It demonstrates that enterprise AI adoption has moved past basic access and chat toward deep workflow integration and delegated agent tasks, establishing new benchmarks for business value.
Topic dossiers OpenAI