Even where firms have moved to modern data platforms, the data itself remains the hard problem. Fragmented sources, inconsistent tagging, missing data definitions and unclear business ownership mean AI outputs cannot be trusted even when the platform is right. The heavy lifting in AI adoption is data preparation, not the AI layer itself: firms need golden sources, clear ownership and accountability models before AI can scale. This panel examines what AI-ready data quality requires: how firms are prioritising which datasets matter most, building governed pipelines and establishing the disciplines that keep data fit for purpose, and what firms that have bolted AI onto ungoverned data are learning.