OpenAI just hit the reset button again. The company announced another sweeping reorganization Friday, this time consolidating ChatGPT and Codex into a single AI agent platform as it pivots to compete in the rapidly evolving autonomous AI space. Company president Greg Brockman now leads all product development in the biggest strategic shift since last month’s leadership shake-up. The move signals OpenAI’s urgency to defend its turf against rivals like Anthropic and Google, who are racing to dominate the agent market that analysts predict will hit $47 billion by 2028.

OpenAI can’t seem to stop reorganizing, and this time the stakes are existential. The company dropped another org chart bomb Friday, consolidating its flagship products and handing Greg Brockman the keys to the entire product kingdom as it races to win what insiders are calling the “agent wars.”

In an internal memo obtained by The Verge, Brockman laid out the strategic rationale with unusual bluntness. “Since OpenAI’s product strategy for this year is to go all-in on AI agents, the company is combining its products to invest in a single agentic platform and to merge ChatGPT and Codex into one unified agentic experience for all,” he wrote. The message is clear – OpenAI is betting the farm on autonomous AI assistants that can actually do things, not just chat about them.

The timing isn’t coincidental. Anthropic has been making serious noise with Claude’s computer-use capabilities, while Google keeps teasing Project Astra’s real-world agent abilities. Microsoft, OpenAI’s biggest backer and partner, is pushing its own Copilot agents across the enterprise. OpenAI found itself with multiple products competing internally while rivals consolidated around single agent platforms.

Brockman’s expanded role represents a major power shift inside the company. The co-founder and president now oversees all product development, a consolidation that comes as OpenAI continues operating under structural changes from last month’s upheaval. That’s when Fidji Simo, who led the AGI readiness division, went on medical leave, triggering a cascade of reporting line changes that apparently haven’t fully settled.

The ChatGPT-Codex merger is particularly revealing about where OpenAI sees the market heading. ChatGPT became the consumer darling with over 200 million weekly users, while Codex powered developer tools and coding assistants. Combining them suggests OpenAI wants to build agents that can both understand natural language requests and execute technical tasks autonomously – exactly what enterprises are demanding as they move beyond chatbots to AI that completes workflows.

“We’re seeing every major AI company pivot to agents right now,” says one former OpenAI engineer who left last quarter. “The question isn’t whether agents are the future, it’s who builds the platform that everyone else builds on top of.” That platform race is driving OpenAI’s restructuring frenzy, even as the constant changes risk creating internal chaos.

The enterprise opportunity is massive and immediate. Gartner predicts that by 2028, AI agents will be involved in 15% of day-to-day work decisions, up from less than 1% today. Companies want AI that can schedule meetings, write code, analyze data, and interface with other software without human handholding. OpenAI’s previous product structure, with separate teams for ChatGPT consumer features and Codex developer tools, wasn’t built for that unified vision.

But the constant reshuffling raises questions about stability. This marks at least the third major organizational change in as many months, following Simo’s departure and earlier shifts in the research division. Employees told The Verge the changes are creating reporting confusion, though most acknowledge the strategic logic behind consolidating around agents. One current staffer said, “It’s disruptive, but we’d rather disrupt ourselves than let Anthropic or Google disrupt us.”

The competitive pressure is intensifying daily. Anthropic just raised another funding round at a $60 billion valuation specifically to accelerate its agent capabilities. Google’s DeepMind division is rumored to be weeks away from launching Astra for enterprise customers. Even startups like Adept and Sierra are attracting hundreds of millions in funding to build specialized AI agents. OpenAI’s window to establish platform dominance is narrowing.

Brockman’s memo didn’t detail the full extent of the organizational changes or which executives will report to him under the new structure. The lack of specificity suggests OpenAI is still working out reporting lines and team compositions, potentially keeping flexibility to make additional adjustments as the agent strategy evolves. What’s certain is that the company sees 2026 as the make-or-break year for establishing agent market leadership.

For developers and enterprises already building on OpenAI’s platform, the consolidation could mean more coherent APIs and better integration between conversational AI and task execution. But it also means potential disruption to existing Codex implementations as OpenAI merges the technical foundations. The company hasn’t announced a timeline for the product integration or what it will mean for current API customers.

The agent wars are just getting started, and OpenAI is reshuffling the deck while the cards are still in play. Whether Brockman’s unified vision can execute faster than rivals remains the multibillion-dollar question hanging over the entire AI industry.

OpenAI’s latest reorganization isn’t just internal housekeeping – it’s a public admission that the AI industry’s center of gravity has shifted from chatbots to autonomous agents. By merging ChatGPT and Codex under Brockman’s unified product vision, the company is making a bet that whoever builds the most capable agent platform will control the next decade of enterprise AI. The constant restructuring might look chaotic from the outside, but it reflects just how fast the competitive landscape is moving. With Anthropic, Google, and Microsoft all sprinting toward the same goal, OpenAI can’t afford to let organizational structure slow down its product velocity. The real test comes in the next few months when we see whether this unified platform actually ships – and whether it delivers the autonomous capabilities that enterprises are demanding.