OpenAI Reveals How AI-Native Companies Strengthen Workflow Operations

According to OpenAI's latest Enterprise Signals data, the top 10% of AI-using companies generate 8.3x more output tokens per active user compared to average companies, a significant increase from 2.6x in January. This gap suggests a deep shift in operational practices, where companies connect agents to company context and tools, delegate more substantive tasks, and make successful workflows repeatable.

Basis develops AI agents for accounting firms and automated its new employee onboarding process using Codex and reusable onboarding skill. This reduced first-day onboarding time from 2 hours to 30 minutes, allowing HR to focus on culture and support rather than repetitive setup. The process is defined as a skill with clear triggers and completion criteria, enhancing consistency and improvability.

Clay built a self-learning revenue engine for GTM teams, introducing account-specific persistent workspace and subagent to manage scattered sales data. Each subagent reviews key sources like CRM, email, and Slack to update deal folders overnight, while a coordinating agent suggests priority actions in the morning. This workflow saves about 1 hour of overnight inbox triage time and helps sales reps avoid missing small follow-ups in long sales cycles.

Exa Labs is pursuing an 'Exa everywhere' strategy to expand the developer ecosystem. Although the text was cut off, they emphasize using agents to connect opportunities to validated actions to provide search APIs wherever developers need them. All three cases demonstrate a common evolutionary stage: training agents as stable processes, providing persistent context for changing work, and converting them into actionable opportunities.