Ep 6 | From prototype to production | Mustafa Ghafouri

An impressive AI demo can make production feel like a formality. But the leap from a controlled sandbox to a system that delivers real value is where most AI initiatives stall. In this episode, we speak with Mustafa Ghafouri - Head of Data Science at Pharmacy2U - about why so few AI prototypes make it to production, and what the ones that succeed get right. Mustafa left his medical degree for data science over a decade ago. Since then, he's consistently shipped models into production across healthcare and tech, including as a solution architect at Databricks and on projects with NHS Digital. He explains why operationalising AI is the most underestimated part of any project, which questions he asks before he believes a demo, and why he builds the solution before the model. We also look at what made an NHS content-moderation project work, what has held back ambitious ideas like AI scribes and automated triage, and why simple use cases done well so often beat grand plans. If you're about to launch your first major AI initiative, or you're stuck somewhere between promising prototype and real-world impact, this is a conversation for you.