Apertus (ETH)
2025 challenges — Zurich
How can we build trustworthy open-source multi-agent AI?
Problem
AI is rapidly shifting from Generative to Agentic, and soon to Physical Intelligence. Today’s systems are mostly proprietary, built by a handful of large companies, leaving little room for openness, trust, or (self-)sovereignty. Without open and reliable alternatives, researchers, entrepreneurs, and public institutions risk dependency. What is missing are multi-agent systems that are not only powerful, but also trustworthy and accessible for broad use.
Why hack?
Shape how the next wave of AI is built: open, trustworthy, and in the hands of many. Work with cutting-edge open-source tools before they reach industry scale. Gain exposure to ETH AI Center’s & ETH Entrepreneur Club's industry networks — from EY, Google, IBM, and SWISS to innovative startups like LatticeFlow. Potential to turn prototypes into impactful solutions, startups, or investment opportunities.