The Python client Pyeapi (http://pyeapi.readthedocs.io/en/master/index.html) library is a native Python library wrapper around eAPI. It provides a set of bindings to configure Arista EOS nodes. Why do we need Pyeapi when we already have eAPI? Picking between Pyeapi versus eAPI is mostly a judgment call if you are in a Python environment. 

However, if you are in a non-Python environment, eAPI is probably the way to go. From our examples, you can see that the only requirement of eAPI is a JSON-RPC capable client. Thus, it is compatible with most programming languages. When I first started out in the field, Perl was the dominant language for scripting and network automation. There are still many enterprises that rely on Perl scripts as their primary automation tool. If you're in a situation where the company has already invested a ton of resources and the code base is in another language than Python, eAPI with JSON-RPC would be a good bet.

However, for those of us who prefer to code in Python, a native Python library means a more natural feeling in writing our code. It certainly makes extending a Python program to support the EOS node easier. It also makes keeping up with the latest changes in Python easier. For example, we can use Python 3 with Pyeapi!