How teams trust agent-written code
We’re studying how engineering teams actually build (and lose) confidence in AI agent-authored code. Your answers will feed a public benchmark report, analyzing these behaviors across the industry, similar in spirit to the DORA or Stack Overflow surveys.
This should take about 15 minutes. There are no right answers, we’re interested in what actually happens on your team, including the messy parts. Individual responses are confidential; only aggregated results are published. A few of the early questions determine whether the rest of the survey applies to you, so please answer honestly even if it means the survey ends early for you.