The post asks about general platforms or best practices for testing workflows. The comments suggest using tools like Arize to log workflow steps and emit debug events. However, the community member is more interested in creating a testing platform to ensure the system behaves as expected, such as never talking about certain topics or responding with specific responses to certain inputs. While Arize is mentioned as having some of these capabilities, there is no explicitly marked answer in the provided information.
Probably using something like arize would help you log your workflow steps. Alternatively, I just add prints (or emit debug events that I use with stream_events )
I'm not looking to emit the events or do monitoring (yet) though. I'm more asking how we might create a testing platform to ensure that the system is always doing what it's supposed to be doing?
Say, we want the platform to never talk about sex, drugs, and rock and roll, or when a person says "X" you should respond with "Y". etc
