In this talk, I examine how example-driven interfaces (EDIs) can support a variety of data tasks. EDIs are a class of mixed-initiative user interfaces that enable users to specify tasks such as data extraction, cleaning, transformation, querying or analysis, in the casual and familiar language of examples. Through tools that we have built and evaluated for data extraction (SEER and Texture), querying (DataPlay and Qetch), generation (Synner) and debugging data pipelines (WhyFlow), I demonstrate that designing such systems brings together new methods from a variety of research fields and opens up exciting research opportunities not only in HCI but also in how we design data management systems.