Cassis

Cassis built a repeatable benchmark combining 30 Formula 1 questions from the Spider 2.0 database with 60 business cases to compare LLM analytics agents, and found Anthropic's Sonnet 5.5 ranked first …

Cassis has released an open-source context bootstrap kit on GitHub to help teams assemble initial context for analytics agents from existing data stacks, replacing a token-heavy agent stage that spent…

A study by Cassis on LLM context optimization for data analytics found that adding a full example library raised Haiku's accuracy from 70% to 82% and Sonnet's from 81% to 93%, while removing rules and…