How frontier firms are pulling ahead

· Source: OpenAI News · Field: Business & Management — Corporate Strategy & Leadership, Operations & Process Management, Project & Product Management · Depth: Intermediate, medium

Summary

OpenAI's B2B Signals research, released May 6, 2026, reveals that "frontier firms" (top 5% of AI users) now utilize 3.5x more intelligence per worker than typical firms, an increase from 2x a year prior. This advantage stems from deeper, more complex AI integration rather than just higher message volume, with only 36% of the gap explained by activity. Agentic workflows are a key differentiator, as frontier firms send 16x more Codex messages per worker. The report introduces B2B Signals as a recurring measure of AI diffusion in businesses, based on de-identified, aggregated enterprise usage data. It highlights that AI use is broadening across functions like IT, software development, and finance, moving beyond general productivity to specialized tasks.

Key takeaway

For AI Architects and Product Managers aiming to maximize enterprise AI value, focus on enabling deep, agentic workflows rather than just broad access. Your strategy should shift from measuring basic adoption to fostering complex, delegated tasks using advanced tools like Codex, and integrating AI directly into core business functions. This approach will help your organization move towards the "frontier" of AI utilization and achieve compounding benefits.

Key insights

Frontier firms gain compounding advantages by integrating AI deeply into complex, agentic workflows.

Principles

Method

OpenAI's B2B Signals uses privacy-preserving, aggregated enterprise usage data to measure AI diffusion, focusing on depth, tool association, and broadening use cases across industries and functions.

In practice

Topics

Best for: AI Architect, AI Product Manager, Entrepreneur, Director of AI/ML, VP of Engineering/Data, Consultant

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by OpenAI News.