Enterprise AI gap widens: top firms now generate 8.3× more tokens

The report from OpenAI highlights a growing divide between companies that effectively integrate AI into their operations and those that use it more casually. The increase in token generation from 2.6× to 8.3× since January is not just about usage volume but reflects a more profound operational change. Frontier firms are moving from using AI for simple assistance to letting it execute substantive tasks by connecting agents to their own company data and tools.

The case studies of Basis, Clay, and Exa Labs illustrate this principle in action. Basis has turned its onboarding process into a reusable skill, dramatically cutting the time required. Clay uses dedicated agents per account to keep deal context current, saving sales staff significant time. Exa Labs automated a workflow for Codex that monitors for integration opportunities and prepares tested code for human review. In each case, the key is not just the automation itself, but building a system with clear triggers, defined 'done' states, and human checkpoints.

The trajectory suggests that the primary challenge for enterprise leaders is not adopting AI, but managing its integration into existing human systems. The success of these initiatives may hinge on defining clear outcomes, assigning ownership, and creating a cycle of experimentation and refinement. The evidence implies that the capacity to scale workflows—not just the technology itself—is likely to be the defining factor for companies seeking to realize value from AI investments. The report gives a structured six-step process for others to follow and emphasizes that early-career employees may be quick adopters (sending 13 more messages per week after six months), a sign that a company's future operating model may be built around this kind of human-AI collaboration.