Two new studies published by OpenAI examine how AI adoption is spreading across firms and workers, and what differentiates organizations that use AI most intensively. The central finding is that enterprise AI is moving from answering questions to carrying out multi-step work: assistants help people think, while agents complete tasks.
The studies
- Enterprise Signals provides a practical look at agentic AI across OpenAI’s enterprise customers, detailing what frontier firms do differently and where agentic work is spreading.
- The companion working paper, How Organizations Use AI: Evidence from ChatGPT, analyzes how adoption grows across companies, roles, and seniority levels.
Key measures and trends
-
Depth of use: Each month OpenAI ranks enterprise customers by output tokens per active user. Frontier firms—the top 10% that month—generated 8.3× as many output tokens per active user as typical firms (45th–55th percentile) in June, up from a 2.6× gap in January. That widening “frontier gap” appears across industries and company sizes.
-
Codex vs ChatGPT output: In June, Codex accounted for 64% of combined Codex and ChatGPT output tokens among enterprise customers. Because agentic, multi-step workflows generally produce more output, this share reflects both increased Codex usage and the larger outputs those workflows create.
-
Organizational correlates: Among U.S. public firms studied in How Organizations Use AI: Evidence from ChatGPT, earlier enterprise adopters held more assets, employed more workers, and invested more in R&D than non-adopters. The reports suggest that access to models alone may be insufficient; complementary investments in continuous employee learning, shared workflows, data infrastructure, and governance support broader adoption.
What makes agents effective
-
Context and tools: Agents need access to the right company context and tools. Plugins bundle reusable skills with app connections to company data and actions, enabling agents to complete specific workflows. For example, a sales Plugin might combine a team’s playbook with CRM access so an agent can draft a tailored response using current customer information and past proposals.
-
Frontier lead on advanced capabilities: Among weekly active users, 21% at frontier firms use Plugins and 19% use skills, versus only 9% and 3% at typical firms. Still, frontier adoption is below what’s possible: OpenAI’s internal usage shows weekly Plugin usage by 95% of active users, illustrating the potential for much deeper adoption of these capabilities.
Where adoption is growing
-
Functional spread: Software engineering was an early center of agentic adoption, but Codex use has grown rapidly across other knowledge-work functions since February: weekly active enterprise Codex users increased 108× in legal, 41× in sales, 41× in recruiting, and 26× in marketing, compared with 5× in engineering.
-
Example: At Virgin Atlantic, the shift is visible across the business. Engineering teams use Codex to refactor legacy code in 30 minutes instead of two weeks; product teams use ChatGPT Work to compress weeks of competitive research into hours, informing the airline’s five-year digital strategy.
Patterns within organizations
- Contrary to many surveys that report higher AI use among leaders, administrative data from millions of conversations show that six months after adoption early-career employees sent 13 more messages per week than executives. For leaders, this indicates a practical opportunity to identify employees with strong AI habits and make their effective workflows visible so practices can spread.
Implications for leaders
Companies may have access to the same models, but frontier firms are deploying them faster and more deeply across their organizations. The practical agenda the reports point to is clear: connect agents to the context and tools needed to complete valuable work; establish permissions, review, and governance; and help employees convert effective individual workflows into shared ways of working. Doing so can help organizations move from assistance to execution.
For more detail, Enterprise Signals explores how frontier firms use AI across industries and functions, and OpenAI enterprise customers can request a customized benchmark to compare against frontier firms. How Organizations Use AI: Evidence from ChatGPT provides evidence on how adoption spreads across firms and workers.
