AI can change your code faster than you can review it.
That's not necessarily a problem.
The interesting question is:
How do you know what that change impacts?
Imagine asking an AI coding agent:
“Refactor this service.”
A few seconds later, it has changed 15 files.
The code compiles.
Tests pass.
The diff looks reasonable.
But then I wonder:
- Which methods depend on the one that changed?
- Which API endpoints eventually reach it?
- Did the change affect something that isn't covered by the tests?
- Is there a caller the AI didn't need to touch, but that still depends on the old behavior?
AI is getting very good at making changes.
I think we're going to need better tooling around understanding the consequences of those changes.
That's one of the problems I'm currently exploring with ImpactCheck.
Curious how other developers handle this:
When an AI agent makes a large code change, how do you check its blast radius before opening the PR?
Top comments (1)
Your insights on the challenge of understanding the impact of AI-driven code changes are spot on. I've faced similar issues when integrating AI tools into our workflow, particularly around ensuring that all dependencies are accounted for before merging changes. One approach that has worked well for me is implementing a dependency graph analysis tool that can visually represent these relationships, helping to identify potential issues. If you’re looking for help refining ImpactCheck or enhancing your tooling in this area, I’d be glad to discuss a paid collaboration. How do you envision the ideal solution for tracking impact?