Create a mechanism that would allow the end-user (human) can provide / critique the model that was auto-selected based on the chat input. Maybe after the response is provided then we could go back to the input we submitted and have a button to click next to the edit button and allow us to select a score 1-5 and allow for free text (optional) to provide the feedback to that particular interaction. Then ingest that into a feedback loop that Hatz team that manages auto-model can use as general feedback to and maybe tighten this going forward.