AI is now embedded in pre-screening and patient matching across a growing number of sites, sponsors, and recruitment vendors. This session looks at what is actually working, what is quietly failing, and why the difference often has nothing to do with the algorithm. Panelists examine the upstream conditions that determine whether matching succeeds at all, including protocol design, inclusion and exclusion criteria written for humans rather than machines, EMR data that was never meant to be queried this way, and recruitment spend that gets committed before anyone checks the feasibility. They also take the conversation past consent, where the real test happens. Attendees will leave with a clearer sense of where AI earns its place in the screening workflow, and where a human still has to make the call.