This Forum was a glimpse into how engineering is changing in real time. Chris, Romain, and Friel didn’t just talk about coding with AI—they showed what it looks like when AI acts as a true teammate, helping engineers plan, build, and debug while people stay firmly in charge of the results. They demonstrated Codex running a long, complex build that not only produced code but also managed sub-agents, tested ideas, and documented its work like a seasoned collaborator. Nearly every OpenAI engineer uses Codex, leading to better reviews, fewer bugs, and faster progress on meaningful projects. But it’s not just for experts—non-technical teams are using it too, asking questions about code or making small changes on their own. These tools aren’t replacing creativity or judgment—they’re freeing people up to focus on what really matters. Codex isn’t just “AI writing code”; it’s a way for teams to work smarter, move faster, and actually enjoy the process again.

The OpenAI Forum hosted a conversation between Stack Overflow CEO Prashanth Chandrasekar and OpenAI’s Head of Developer Experience Romain Huet on how AI is transforming software work and the role of developers. Prashanth reflected on his journey from building early hospital management software in C++ to leading Stack Overflow and returning to hands-on coding with modern AI tools, framing generative AI as a platform shift on the scale of the internet. He argued that AI is not replacing developers but changing what it means to be one, creating effectively infinite demand for code and new applications across domains like drug discovery, while increasing the need for engineers who understand fundamentals and can guide, evaluate, and collaborate with AI agents. The discussion highlighted how enterprises are piloting and scaling AI, navigating security, privacy, and governance concerns, and using products like Stack Overflow for Teams plus OpenAI models to power internal assistants that tap institutional knowledge. Prashanth also described how Stack Overflow is evolving—introducing AI Assist, staging areas for questions, opinion-based discussions, and chat—to make learning more empathetic while preserving high-quality, trusted knowledge amid rising AI usage but uneven trust. He closed by advising engineers and universities alike to pair deep foundational skills with mastery of AI tools, emphasizing that those who can do both will be especially valuable as AI lowers the barrier to innovation and accelerates progress in high-impact fields.