In the realm of AI-assisted programming, software engineers show a remarkable openness to adopting new tools. Recent insights reveal that among the hundreds of engineers at Notion—the company behind the widely-used task management software—the adoption growth rates of Anthropic’s Claude Code and OpenAI’s Codex have outstripped that of Cursor. The internal shifts at Notion mirror a broader industry trend: engineers are transitioning from AI tools integrated within Integrated Development Environments (IDEs) to programming agents capable of autonomously completing comprehensive tasks, which can also be leveraged by non-technical users.

Although Cursor was early to introduce programming agent features, its brand identity remains closely associated with IDE assistance tools. In contrast, Claude Code and Codex have gained recognition for their autonomous programming prowess, excelling in tasks such as writing new features from the ground up or refactoring existing codebases.

At Notion, Claude Code is more frequently used by junior engineers or for simpler tasks, thanks to its adeptness at anticipating user intent. On the other hand, senior engineers tend to favor Codex for tackling complex, time-intensive projects. While engineers may employ multiple tools depending on the task at hand, the rising adoption rates of Claude Code and Codex suggest they are steadily becoming the tools of choice for Notion’s engineering team.