← All speakers
Julius Brussee builds infrastructure for cheaper, more disciplined AI agents while studying Data Science and AI at Leiden University. He created Caveman, an open-source coding-agent skill with more than 90,000 GitHub stars, then expanded the idea into tools for context compression, spend measurement, and eval-gated optimization. His other projects include Cavekit, a spec-driven agent build loop, and Cavemem, a local-first memory layer shared across coding agents. Julius has spoken at Prosus AI House, DEPT’s agentic show-and-tell, The AI Meetup, and BNR’s De Grote Tech Show.
