As AI agents become increasingly capable of writing software, a debate is emerging. Do we still need to read and review all the code they produce?

Code reviews and pull requests have long been fundamental to how we build software. They help us catch issues, share knowledge, and build confidence in the changes we make. But as agents produce more and more code, is reading every line really the best way to establish that confidence?

In this talk, we’ll take a step back and explore why we review code, what we’re trying to achieve, and whether there are better ways to build trust in the software our agents produce. You’ll leave with practical examples and takeaways you can apply to your own team.

Now that AI has changed how we write software, perhaps it’s time to question how we review it too. To read, or not to read? That is the question.

As AI agents become increasingly capable of writing software, a debate is emerging. Do we still need to read and review all the code they produce?

Code reviews and pull requests have long been fundamental to how we build software. They help us catch issues, share knowledge, and build confidence in the changes we make. But as agents produce more and more code, is reading every line really the best way to establish that confidence?

In this talk, we’ll take a step back and explore why we review code, what we’re trying to achieve, and whether there are better ways to build trust in the software our agents produce. You’ll leave with practical examples and takeaways you can apply to your own team.

Now that AI has changed how we write software, perhaps it’s time to question how we review it too. To read, or not to read? That is the question.