AI agents are everywhere. Last June, 31% of you said you were using them. Since then, they’ve graduated from auto-completing snippets to handling the heavy lifting, sometimes even building entire software tools from the ground up. In fact, research shows that PRs handled by agents and humans together often have better testing than those done by humans alone. We’re not hurt; we’re just taking notes.
The way we work is changing. You’re moving from being the person who writes every line to the one who edits and governs the output. We want to know where you sit on the spectrum of autonomy. Are you still "human-in-the-loop," or have you moved to "human-on-the-loop," overseeing a bot that’s doing the driving?
As part of our research into how software engineers work, we’re fielding a new survey on agent use in the workplace. We’ve asked about AI use, but the way AI works with us has changed, so we want to know if your work has changed too. Do you let agents run autonomously? Do you pay per token or flat rates? Most importantly, are agents helping?
Chime in and help us define the new limits of responsibility. It only takes a few minutes, and unlike your agent, we won’t try to refactor your entire codebase while you’re at it.
This survey will be open until April 30. In May we’ll share the results.