OpenAI reports 97.9% employee adoption of Codex as agentic AI usage surges across departments

OpenAI Claims Near-Universal Codex Adoption Among Employees

OpenAI has published research showing that 97.9% of its employees now use Codex, its AI coding and work agent platform, marking a dramatic increase from roughly 40% in August 20251

1

. The company's paper, titled "The Shift to Agentic AI: Evidence from Codex," presents this internal adoption as evidence of a fundamental transformation in how workers interact with AI tools. According to OpenAI, Codex now accounts for 99.8% of output tokens generated across both Codex and ChatGPT among its employees4

4

. Every department within the company, including non-technical divisions such as the legal department and recruiting department, now relies on Codex as their primary AI tool for work2

2

. However, all these metrics come from self-reported data by OpenAI itself, a company with direct financial incentive to promote the product it is measuring3

3

.

The Shift from Chatbots to Agents Accelerates

The research documents what OpenAI characterizes as a market-wide transition from conversational chatbot interactions to autonomous agents that execute multi-step tasks. Instead of one-off ChatGPT prompts seeking advice or information, workers are now assigning Codex agents to tackle complex, long-running work that can span hours1

1

. The number of active users has grown more than fivefold in the first half of 2026, with the most rapid increase occurring outside the initial audience of software developers2

2

. Since the start of the year, the share of individual Codex users who submit at least one request for a task estimated to require more than eight hours for an experienced human to complete has increased nearly tenfold1

1

. This shift to agentic AI represents a fundamental change in workplace AI interaction, moving from advisory tools to systems that perform actual knowledge work automation.

Non-Developer AI Usage Surges Across Organizations

The most striking growth appears among non-technical users. Non-developer AI usage of Codex has risen 137x for individuals, 189x for organizational users, and 12x within OpenAI since August 20252

2

. In June 2026, the median OpenAI employee in a legal role generated 13 times more monthly output tokens across Codex and ChatGPT than they did in November 2025, while researchers saw their output increase over 50 times5

5

. Among a sample of individual Codex users, 80.6% made at least one request estimated to represent more than 30 minutes of work by an experienced human, while 70.2% delegated work estimated to save more than an hour, and 25.6% had assigned tasks estimated to take more than eight hours4

4

. The platform now boasts over 5 million weekly active users, a more than sixfold increase since February5

5

.

External AI Agent Adoption Lags Behind Internal Metrics

While OpenAI's internal adoption appears nearly universal, external organizational adoption tells a different story. Among active users of ChatGPT and Codex at organizations outside OpenAI, just above 0% used Codex in August 2025, but that share has now climbed to around 17.3%4

4

. Individual adoption remains minimal at approximately 0.7%1

1

. This gap between internal and external adoption raises questions about how representative OpenAI's own workforce is of typical enterprise users. The company has cautioned that its internal data may not portray typical trends of enterprise adoption, citing favorable conditions like high organizational commitment and unrestricted usage5

5

. OpenAI did not clarify whether it incentivizes or encourages employees to use its AI tools through internal communications, token allocations, token usage leaderboards, or by tying tool usage to performance metrics1

1

.

Questions Around Self-Reported Productivity Gains

The paper frames increased token usage and longer task requests as evidence that agents are handling more complex work and delivering employee productivity gains. However, no independent third party has verified any of the usage figures3

3

. The productivity claims also face scrutiny because faster code generation does not automatically translate into proportional productivity gains—verification, testing, and deployment time may expand to absorb the speed improvement2

2

. The paper does not present data on whether the shift to agents has measurably improved output quality or reduced total time to completion3

3

. Workplace culture expert Jessica Kriegel notes that agents are "reducing what I'd call the psychological cost of action," making unfamiliar work feel more approachable, which means users "start sooner, experiment more, and spend less energy worrying about what I don't know"4

4

.

Competitive Pressure and Market Positioning

The broader context is an intense race among AI companies to prove that agents, not chatbots, represent the next phase of the market. OpenAI merged ChatGPT and Codex under Greg Brockman in May, consolidating its product strategy around a single agentic platform ahead of a potential Q4 IPO3

3

. Competitors including Anthropic's Claude Code and Google Gemini are pursuing similar agentic strategies, making the competitive pressure to demonstrate AI agent adoption growth intense3

3

. The company expanded Codex earlier this month with enterprise plugins connecting 62 business applications, positioning it as mainstream workplace tools for knowledge workers3

3

. OpenAI insists that longer running tasks consume more tokens, and to the extent those can be billed, that should help address hundreds of billions in debt obligations1

1

. The paper is most useful as a signal of where OpenAI believes the market is heading, and where it wants investors to believe the market is heading before its IPO, though whether adoption is happening at the pace and scale that self-reported data suggests remains a question that only independent measurement will answer3

3

.