Parallel Reality

Select a failure scenario Each scenario contains the full conversation history leading up to an AI response that Leonard disliked.

You can edit the AI's response (or any part of the conversation) and re-send to the LLM to test behavioral changes.