Governing AI Quality:

How to Move from Output to Outcome

Most AI in localization conversations are stuck on speed and cost. The teams getting real value from AI have moved past that question. They're asking a harder one: how do we know what we're shipping is good, and how do we make it better with every project? That's a governance problem, and most localization programs don't have the infrastructure to answer it. In this session, Imran Sadiq, Hilary Wright, and Leah Wanta walk through the layer that turns AI output from a black box into something measurable, auditable, and improvable over time. They'll introduce COMET, an automated quality scoring framework purpose-built for translation, and show how it fits into a broader quality loop that makes AI output defensible to stakeholders and better with every project.