Agents & Evals: Berlin Meetup

​Building AI systems that actually work—over talks, snacks, and good conversation.

​We’re gathering folks working on LLMs, agents, and evals to share what we’re learning and swap notes on what’s working in the real world. Two short talks, snacks, and time to hang out.

​🎤 Talks

​🔧 Why Tool Calling Breaks Your AI Agents—and What to Do Instead

​Alan Nichol – Co-founder & CTO, Rasa
Agent frameworks often throw tools at LLMs and hope for the best. Alan will walk through how Process Calling helps systems handle real user requests with memory, branching, and control.

​🧪 Escaping POC Purgatory with Evaluation-Driven Development

​Hugo Bowne-Anderson – Vanishing Gradients
Most LLM apps stall after the demo. Hugo will share lessons from teams who escaped that trap—using logging, eval harnesses, and synthetic data to build systems that actually hold up.

​⚡ Lightning Talks

​​Show us what you’ve built. We’ll open the floor for 5-minute lightning demos—tools, workflows, prototypes—then stick around to chat and trade ideas.

​​We’ll circulate a form closer to the date so people can sign up to give lightning talks. In the meantime, if you're interested, please do reach out.

​​We’re excited to connect with the Berlin AI/dev community, see what you’re building, and hopefully find people to build with in the future.

​We’re excited to chat with others building in this space—whether you're deep into evals, just exploring agents, or somewhere in between.

​

​Meeup will be in Prenzlauerberg: address will be provided after registration.