We are in the era of in-IDE coding agents. Tools like GitHub Copilot, Claude Code, and Cursor are obvious winners in this space. They have made AI-assisted development mainstream and have become part of many developers’ daily workflows. However, there is a scalability issue with all three of them.The workflow currently followed while coding with AI agents usually looks like this: a ticket is assigned to a human developer who has access to these tools. The developer opens the IDE and uses the agent mode to complete the task. Very often the agent gets most of the code right, but if not, the developer fixes the remaining parts and the task is finished.The problem is not at the beginning or the end of this workflow — the problem lies in the middle.Companies that are paying for these tools will eventually analyze their usage. At some point they will look at how much production code is actually being written by these agents. Once organizations start seeing those numbers, they will…
No comments yet. Log in to reply on the Fediverse. Comments will appear here.