A few notable data choices for the LLM hackday: bonded pairs (Pepper + Patch), a trio of guinea pigs, an African Grey Parrot with 200-word vocabulary, and a 25-year-old tortoise with CITES documentation — good variety for testing LLM reasoning ("which animals require expert experience?")