The community members discuss the relationship between pydantic or guidance and raw data. The original post suggests that these tools are not directly connected to raw data, but rather create something from a class that is used as the data. One community member agrees with this understanding, stating that it's about getting the language model to output a certain structure. However, another community member points out that you can connect these tools to your data, providing the example of a guidance output parser that works on top of an index.
