OpenAI B2B Signals: How Frontier Firms are Pulling Ahead
OpenAI B2B Signals: How Frontier Firms are Pulling Ahead
TL;DR
OpenAI has introduced B2B Signals, a business extension of OpenAI Signals that uses privacy-preserving, aggregated enterprise data to track how AI is diffusing across businesses. The primary finding is that "frontier firms"—those in the 95th percentile of AI usage—are pulling ahead of typical firms by using AI for more complex, delegated work rather than simple query-response interactions.
The Compounding Frontier Advantage
Frontier firms are achieving a significant lead in AI utilization, with a gap that is widening over time. As of May 2026, frontier firms use 3.5x as much intelligence per worker as typical firms, an increase from 2x in April 2025.
This advantage is not driven by simple activity volume. Message volume only explains 36% of this gap; the majority of the advantage comes from depth of use. OpenAI uses tokens generated as a proxy for "intelligence demanded," indicating that workers at frontier firms are providing richer context and asking AI to execute more substantive, complex work rather than simply answering questions.
The Shift Toward Agentic Workflows
AI maturity is increasingly defined by the transition from chat-based assistance to the delegation of multi-step tasks to AI agents.
Frontier firms show a disproportionate advantage in the adoption of advanced and agentic tools. The most significant gap is seen in Codex, where frontier firms send 16x as many messages per worker as typical firms. Similar patterns are observed with ChatGPT Agent, Apps in ChatGPT, Deep Research, and GPTs.
Case Study: Cisco
Cisco utilizes Codex to optimize software work across its engineering organization. By treating Codex as "part of the team," Cisco reported the following gains:
- Build times: Reduced by approximately 20%.
- Engineering hours: Over 1,500 hours saved per month.
- Defect-resolution throughput: Increased by 10-15x.
Specialized and Production AI Use
While AI adoption is broad in writing and communication, it is becoming increasingly specialized by business function:
- IT and Security: Focus on how-to and procedural guidance.
- Software Development and Data Science: High concentration of coding usage.
- Finance: Focused on analysis and calculation.
Beyond general productivity, companies are deploying AI via APIs into in-app assistants, developer tools, and customer support systems.
Case Study: Travelers Insurance
Travelers Insurance developed an AI Claim Assistant using OpenAI technology to guide customers through the first notice of loss, answer policy questions, and create claims directly within their internal systems. The company expects the assistant to handle approximately 100,000 calls in its first year.
Strategies for Moving Toward the Frontier
OpenAI identifies that the gap between typical and typical firms is not fixed. Leading firms often treat AI enablement as core infrastructure and use AI to help employees build the necessary skills and habits.
To move toward the frontier, organizations are encouraged to:
- Measure depth of use rather than just seat count or access.
- Build governance that enables production-level deployment.
- Invest in enablement and education to help employees build confidence.
- Identify and scale the impact of frontier teams within the organization.
- Transition from chat toward delegated work using agents.