Event-sourced systems thrive on strict boundaries, deterministic rules, and crystal-clear histories. Large language models thrive on ambiguity, messy inputs, and human-like reasoning. Oliver will show how the two worlds collide - in a good way. We’ll explore natural-language command suggestions that never bypass validation, projections enriched with summaries and insights, event-stream analysis that explains how outcomes were reached, and LLMs advising sagas while domain logic stays firmly in charge. Expect practical demos and lots of source code!