ResearchOpenAI

OpenAI B2B Signals: Frontier Firms Use 3.5x More AI per Worker

OpenAI's B2B Signals report reveals frontier firms use 3.5x more AI per worker than peers, with 16x more agentic workflows. What the data means for CI teams.

7 min readPublished 2026-05-12

What happened

OpenAI published a new research initiative called B2B Signals, designed to track how enterprise AI adoption is evolving across industries and functions. The inaugural report compares "frontier firms" — organizations at the 95th percentile of AI usage — against typical enterprises, and the gap it reveals is substantial and widening.

The headline finding: frontier firms now use 3.5 times as much AI intelligence per worker as typical firms, up from 2x a year ago. This is not a measure of how many employees have access to AI tools — access has become broadly distributed. Instead, the metric captures depth: the complexity of prompts, the richness of context provided, the sophistication of outputs generated, and the extent to which AI is embedded into daily workflows rather than used for occasional one-off tasks.

The report introduces several additional metrics that quantify the adoption gap. Message volume — the simplest measure of usage — explains only 36 percent of the frontier advantage. The majority of the gap comes from what OpenAI terms "richer, more complex AI use." Workers at frontier firms provide more detailed context, request more substantive analysis, and integrate AI outputs into multi-step workflows rather than treating it as a search replacement. Critically, the report highlights agentic AI as the sharpest differentiator: frontier firms send 16 times as many Codex messages per worker as typical firms, suggesting that the most advanced organizations are already delegating meaningful work to AI agents rather than merely using AI for assistance.

Why it matters for practitioners

For competitive intelligence teams, OpenAI's B2B Signals data reframes AI adoption from an operational efficiency initiative to a measurable competitive advantage. The data provides, for the first time, a quantitative benchmark that CI practitioners can use to assess where their organization — and their competitors — stand on the AI adoption curve.

1. The adoption gap is accelerating, not stabilizing. The 3.5x gap, up from 2x a year ago, means frontier firms are pulling away from their peers at an increasing rate. This is consistent with technology adoption patterns where early leaders compound their advantage through better data, more refined workflows, and organizational learning that latecomers cannot shortcut. For CI teams, this acceleration means the window for competitive response is narrowing. Organizations that treat AI adoption as a 2027 initiative are already behind, and the gap will be harder to close with each passing quarter.

2. Depth, not access, is the real differentiator. The finding that message volume explains only 36 percent of the frontier advantage undermines a common organizational assumption: that making AI tools available is sufficient. Frontier firms do not just use AI more frequently — they use it differently. Their workers provide richer context, request more complex analysis, and generate outputs that feed into downstream workflows. For CI practitioners benchmarking their organization's AI maturity, the implication is clear: counting licenses or measuring login rates tells you almost nothing about competitive positioning. The meaningful metrics are workflow integration depth, prompt sophistication, and the percentage of work that is delegated to rather than merely assisted by AI.

3. Agentic workflows are the frontier marker. The 16x gap in Codex usage between frontier and typical firms is the report's most striking data point. It suggests that the next wave of competitive moat building is not about chat-based AI assistance — which is rapidly commoditizing — but about agentic AI systems that can autonomously execute multi-step workflows. For CI teams specifically, this has direct operational implications. Organizations that have moved from manually monitoring competitor activity to automating competitive intelligence with agentic workflows are operating on a fundamentally different efficiency curve than those still relying on analyst-driven processes.

4. The data creates a new competitive benchmarking framework. OpenAI's B2B Signals provides a structured way to think about AI adoption as a competitive variable. CI teams can use the frontier/typical framework to assess competitors: which competitors are likely frontier firms based on their product velocity, hiring patterns, and public statements about AI integration? Which are lagging? This analysis becomes particularly valuable in industries where AI adoption depth directly correlates with operational efficiency, product development speed, or customer experience quality.

Key details

  • Research name: OpenAI B2B Signals
  • Core metric: Intelligence per worker (depth of AI usage, not just frequency)
  • Frontier firm definition: 95th percentile of enterprise AI usage
  • Intelligence gap: 3.5x between frontier and typical firms (up from 2x one year ago)
  • Volume vs. depth: Message volume explains only 36% of the frontier advantage
  • Agentic gap: Frontier firms send 16x as many Codex messages per worker as typical firms
  • Advanced tools: ChatGPT Agent, Apps in ChatGPT, Deep Research, and GPTs all show similar directional patterns favoring frontier firms
  • Key shift: From "how many people have access" to "where AI is deepening workflows and changing how teams operate"
  • Recommendation for organizations: Measure depth, build governance for production use, invest in enablement, scale what works, move from chat to delegated agent work

Market implications

OpenAI's B2B Signals report has implications that extend well beyond the company's own product ecosystem. By quantifying the AI adoption gap with this level of specificity, OpenAI has created a benchmark that will influence enterprise AI purchasing decisions, board-level conversations about digital transformation, and the competitive strategies of technology vendors across the B2B landscape.

For the competitive intelligence industry specifically, the report validates a thesis that many CI leaders have been articulating informally: that AI adoption is not merely an operational variable but a strategic differentiator that belongs in every competitive assessment. When frontier firms are operating with 3.5x more AI intelligence per worker and 16x more agentic workflow activity, the performance differential in areas like market analysis speed, insight generation quality, and competitive response time becomes material. CI teams that can quantify these gaps for their executive stakeholders — using OpenAI's framework or similar benchmarks — will be better positioned to secure investment in their own AI-enabled capabilities.

The agentic workflow data is particularly significant for CI tool vendors. The 16x gap in Codex usage suggests that the next generation of competitive intelligence platforms will need to support agent-based architectures where AI systems autonomously monitor competitors, synthesize intelligence, and deliver actionable briefings — not just provide dashboards and alerts. Vendors that are building toward agentic CI capabilities, such as autonomous competitor monitoring, automated battlecard updates, and real-time deal intelligence, are aligned with the direction frontier firms are already heading. Those still selling primarily analyst-augmentation tools risk falling behind the adoption curve their own customers are navigating.

For practitioners, the immediate application is to use OpenAI's frontier framework as a lens for competitive analysis. Map your key competitors along the frontier-to-typical spectrum based on observable signals: their public AI strategy statements, engineering blog posts, product release velocity, job postings for AI-related roles, and patent filings. Then assess your own organization's position with equal rigor. The 3.5x intelligence gap is not abstract — it manifests in concrete competitive outcomes that CI teams are uniquely positioned to track and quantify.

Related resources