
I like that you can model whatever process you want, and that the fan in / fan out features allow you to run networks of processes in parallel in a predictable way. The dependency resolution allows you to run repeatable builds and reduce the amount of re-work you pipeline has to do. Review collected by and hosted on G2.com.
The server component is not architected to scale across multiple servers, meaning it can be quite slow at a very large scale. Configuration is tricky to manage, as it is not yet possible to just check in a file describing your build process like you can with Travis for example. Review collected by and hosted on G2.com.