Three Claude events in Berlin: a build day tomorrow, then science in November
Dear community,
I am not sure if I've shared this news already - I became Claude Ambassador. Many of you might be subscribed to this newsletter because you took the agentic AI workshop with Xemantic, teaching how to build harnesses on top of the Claude platform. Now I can devote even more time to focus on all the communities I am connected to - AI researchers and scientists, software creators, creatives and artists. The next few weeks in the Berlin Claude community are busy. Here's what's coming up.
Tomorrow: Claude Code Impact Lab
Saturday, 17 October · 10:00–18:00 · Taxfix SE, Köpenicker Str. 122
Tomorrow about 100 people spend a full day building on the themes and problems that came out of our earlier Claude Conversation. We picked participants on purpose: engineers, designers, researchers, founders, creatives and domain experts, put into mixed teams. Teams get Claude API credits and judges, the winning team gets gifts, and the best builds go to Anthropic.
Registration is closed. If you're coming, see you tomorrow. If not, I'll share the highlights in the next issue.
Can AI help us solve science's hardest problems?
Claude Conversation x Foresight Institute Wednesday, 4 November · 18:00–21:00 · CIC Berlin, Lohmühlenstraße 65
AI can now read the literature, write and run analysis code, and propose experiments. Some research groups already run agents in a loop on their open problems. Others still see mostly confident mistakes. This evening is about the space between those two views.
We're bringing together scientists from life sciences, medicine, chemistry and materials, physics, mathematics, nanotechnology, and climate and earth sciences with engineers who build AI agents. It isn't a lecture or a product demo. Most of the evening is small-group discussion around four questions:
- How will AI change your field, and where does it work or fail?
- What would it take: data, tools, skills, verification, trust?
- Which open problems are ready for an AI research loop today?
- Which concrete problems should we build on?
We'll finish with a shortlist of open research problems, which becomes the brief for the Impact Lab three days later.
Entry is free and participants are selected by application. Bring a problem from your own research.
AI Research Loops for Open Science Problems
Claude Impact Lab x Foresight Institute Saturday, 7 November · 10:00–20:00 · CIC Berlin, Lohmühlenstraße 65
On this day we build on the problems scientists name at the Conversation. Teams form around each problem and build a research loop: an agent that reads the literature, writes and runs code, checks its own results, and keeps working. Experienced mentors from the Claude Community support every team.
A loop works best when the scientist who owns the problem sits next to the engineer building the harness, so both kinds of people are welcome. If you're a scientist, no programming experience is needed.
- Morning: short intro, example research loops, and a presentation of Claude Science
- Day: 6–8 hours of building with mentor support
- Evening: team demos (findings, sticking points, next steps) and a panel picks the winner
You get: an open-source research loop on GitHub, $100 in Anthropic API credits per participant, and one month of Claude Max 20x for each member of the winning team.
Bring: a laptop (8 GB RAM, Python 3.10+) and a console.anthropic.com account. Scientists, bring a short problem brief and any public or anonymized data.
Entry is free and participants are selected by application.
The two November events go together. Ideally you come to both: name the problem on Wednesday, build on it Saturday. Both are co-hosted with the Foresight Institute Berlin AI Node, Bradley C Royes and Keith Patarroyo, and continue Foresight's Existential Hope work.
Know a researcher with a problem that keeps them up at night? Forward them this email.
See you soon,
Kazik
