AI Code review

In the company we are constantly concerned about the quality of the code written by our developers. Over time, we have developed specific approaches to organizing code review processes.

Firstly: gitlab continuous integration, in the framework of which we check the quality of code formatting or its compliance to given js or PHP standards.

Secondly: code review processes by senior developers in the form of comments to the submitted merge request.

But if the first part was automated five years ago, it is not that easy with the check of code by a human. Although the company maintains a high level of standardization of the development process, each tech lead is an individual with his own vision of “quality code”.

Recently, we became interested in the possibility of automating Code Review using artificial intelligence. Such a solution can be integrated into modern CI systems and notify developers about problematic parts of the code. This process may not completely replace the human-made code review process, but it can be used for pre-review and to reduce the time of subsequent human review.

It turned out that some of our customers have already advanced in this area, so we are waiting for a new startup that will help make the code better.