Spent over 2 hours going through the Jev docs and this is what i found

Jev is not here to replace your primary LLM model like Claude and ChatGPT; instead, it's best used to make decisions between the outputs you specify. You give it a few choices to pick between along with come context as to what question you're trying to answer with your specified outputs, it then gives you a choice, score, or noul that you would then trigger some logic based off of. I followed the quick start tutorials in this video and showcase how to use it in the TypeSafe Playground, with a curl request, the Python SDK, and with an agent (Claude Code): https://youtu.be/w3Ks3ezL3CQ