Pipeline analyses to find transient failure
For CE/EE pipelines, we suffer from 'random' failures. Either the test could be written better, or the ordering of tests break a certain test. Now keeping track of this is manual labour. But given GitLab has access to both your code and the pipelines it might be able to analyse pipelines by itself and thus maybe find 'bad' tests.
Links / references
None, but at the time of writing we as an organisation struggle to keep master green
By default, 'Pipeline Analyses' is enabled for all pipelines. GitLab will analyse pipelines pipelines of the current project, but also multi project pipelines in which the project is included. If common failures, or patterns in failures are detected GitLab will open an issue on the project informing you of its findings.