Claude Impact Lab x Foresight Institute — AI Research Loops for Open Science Problems

A full-day build at CIC Berlin where domain scientists and AI engineers sit side by side and try to turn a real open research problem into a working agent. The problems come from the Claude Conversation held in Berlin earlier in the autumn, where local scientists named what they most want AI help with. Teams form around those, then spend six to eight hours building a research loop: an agent that reads the literature, writes and runs code, checks its own results and keeps going. Claude Community mentors work alongside the teams, and every team demos at the end. Free, but places are by application and the hosts select participants. Bring a laptop and a console account; teams leave with an open-source loop on GitHub. Why go: most AI-for-science events are talks about the idea. This one puts the scientist who owns the problem next to the engineer who builds the harness for a day, which is the part that usually never happens.